Uploaded by あなたから

The Pollyanna hypothesis

advertisement
(1969)
JOURNAL OF VERBAL LEARNINGAND VERBALBEHAVIOR8, 1-8
The Pollyanna Hypothesis
JERRY BOUCHER 1 AND CHARLES E . OSGOOD
University of Illinois, Urbana, Illinois 61801
The Pollyanna Hypothesis asserts that there is a universal human tendency to use evaluatively positive words (E+) more frequently and diversely than evalua.tivelynegative words
(E-) in communicating. Drawing on existing cross-cultural and developmental data, it was
demonstrated that (a) across a sample of 13 language/culture communities E+ members of
evaluative scales are used significantlymore frequently and diversely than their E - opposites,
(b) across 11 of these communities negative affixesare applied significantly more often to the
E+ members of pairs (to make the E - opposite) than to the E - members (to make the E+
opposite), and (c) across age levels from 7 through 11 E+ members of evaluative pairs appear
earlier, have higher frequencies and diversities of usage and take the negative affix more frequently than their E - opposites. Possible biases in the data and alternatives to the Pollyanna
hypothesis are considered.
In this paper we present some evidence,
drawn from already existing data, for what
we call "the Pollyanna H y p o t h e s i s " - - t h a t
there is a universal h u m a n tendency to use
evaluatively positive (E+) words more frequently, diversely and facilely than evaluatively negative ( E - ) words. Put even more
simply, humans tend to " l o o k on (and talk
about) the bright side of life." This general
hypothesis generates a large number of more
specific and testable hypotheses, e.g., lexicons
will include more E+ than E - terms; for words
matched on other grounds, E+ items will be
used more frequently than E - items; E+ words
will appear earlier in the vocabularies of
children than matched E - words; other things
equal, E+ words will be easier to learn and recall than E - words; other things equal, again,
E+ words will have lower recognition thresholds than E - words; and so on.
F r o m a rather large literature bearing on the
Pollyanna Hypothesis, we select for illustration here only a small, but representative,
group of particularly relevant studies. Using a
continuous association procedure, Bousfield
and Sedgewick (1944) were able to demonstrate
1 The first author is presently at the University of
California Medical Center, San Francisco, California.
1
© 1969 by Academic Press Inc.
that the rate of associating from a restricted
category is a function of the number of items
remaining in the category; subsequently,
Bousfield (1944) demonstrated that both the
rate and asymptote of the function for
Pleasant words was higher than for Unpleasant words, implying that there is a larger
" p o o l " of Pleasant words available. Johnson,
Frincke, and Martin (1961) reported that E+
words, even when matched in frequency of
usage (Thorndike-Lorge) with E - words, are
more "meaningful" (Noble's m ) - - t h a t is, are
more productive of different associations in a
limited time. The same authors found that,
with words matched in both frequency of
usage and m, E+ words had lower visual duration thresholds than E - words.
With respect to learning and retention,
Anisfeld and Lambert (1966) found that E+
words were learned faster than E - words in a
NS-syllable/word paired-association task (although not in word/NS-syllable or word/word
tasks), and they conclude that positive affect
facilitates the acquisition of new meanings
(here, of NS-syllables). Comparing the incidental learning of lists of nouns judged to be
neutral, E+ and E - , all matched on ThorndikeLorge frequency of usage, Amster (1964)
1
2
BOUCHER AND OSGOOD
found E+ to be recalled better than E - and
both affectively toned lists to be recalled
better than neutral nouns. Yavuz (1963) had
Ss learn the purported English translations of
18 Turkish words, six "translations" being
E+, six E-, and six neutral in evaluation; a
week later, the Ss were asked to recall these
English translations, and it was found the E meanings were most poorly recalled.
The most extensive review of literature bearing on the Pollyanna Hypothesis is to be found
in a recent monograph by Zajonc (1968). This
monograph includes a number of new studies
of his own relating the evaluative preference of
words to their Thorndike-Lorge frequencies of
usage--a sample of 154 antonym pairs (of
various form classes) in English, similar data
for French, German and Spanish, and smaller
samples of names of members of various categories (occupations, countries, cities, trees,
fruits, vegetables, and flowers)--and in all
cases preference is found to correlate positively with frequency of usage. On the basis of
additional experimental data relating frequency of mere exposure to evaluation, Zajonc
attributes the causal direction of the wordfrequency--word-value correlations to an
effect of frequency of exposure upon evaluation rather than the reverse--the reverse being
the Pollyanna Hypothesis. We will consider
this question in the Discussion section.
In connection with cross-cultural studies of
affective meaning systems (Osgood, 1964), 100
concepts (nouns) were rated against 50 qualifier scales (adjectives) by native speakers of
more than 15 different languages. When, after
factor analysis, these 100 translation-equivalent nouns were assigned factor scores on the
dominant E factor, it was found that E+ concepts were much more frequent than E concepts in every language group, the average
ratio being 4 to 1 in favor of E+. Since these
concepts were selected on the basis of translation fidelity from an original list of over 200
nouns, drawn with no evaluative malice aforethought from lists compiled by linguists
interested in glottochronology (language
change over time) and from the Human Relations Area Files, it would appear that this
evaluative bias is inherent in human languages
rather than being a happenstance of this particular sampling. However, the latter alternative cannot be ruled out absolutely.
The present paper utilizes extensive data on
qualifier usage drawn both from this crosscultural project and from parallel research by
DiVesta (1964, 1965, 1966) on semantic
development in children, with age levels rather
than language/culture communities being
compared. These data are used to test three
specific predictions deriving from the general
Pollyanna Hypothesis: (a) That across a wide
range of languages and cultures, E+ qualifiers
will have higher frequencies and diversities of
usage than their E - opposites; (b) that negative affixes (e.g., un-, non-, dis- in English) will
be added to E+ qualifiers to make E - opposites more frequently than to E - qualifiers to
make E+ opposites"' (c) that in the
develop/
ment of language in children E+ qualifiers will
appear earlier than their E - opposites and be
more frequent and diverse in usage at all age
levels.
ANALYSISI : FREQUENCYAND DIVERSITYOF
E+ AND E - ACROSSLANGUAGES
Pollyannaism is hypothesized to be a human
universal. All human languages have means of
qualifying (adjectives) substantives (nouns)
and all languages of which we are aware
utilize the principle of opposition in structuring their modes of qualifying. The Pollyanna
Hypothesis leads to the prediction that E+
qualifiers will be used more frequently than
their E - opposites; it also predicts that E+
qualifiers will display greater diversities of
usage, i.e., will be used to modify a wider range
of different nouns, than their E - opposites.
Data being collected in some 25 language/
culture communities lend themselves to the
test of these predictions. The actual analysis is
limited to the 13 communities with data available at the time of this report.
THE POLLYANNAHYPOTHESIS
Method
Since the data-collection process in the cross-cultural
project has been described in detail elsewhere (Osgood,
1964), only those aspects most relevant to the present
analysis will be presented here. A set of 100 culturecommon substantives (e.g., HAND, SKY, SYMPATHY, DEFEAT, MOTHER, etc.) was translated
into each of the languages under study. These were presented to 100 high-school level boys as stimuli in a
restricted word-association task, Ss being instructed
to give the first qualifier that occurred to them for each
substantive, e.g., TREE-green, HOUSE-big, GIRLpretty in English. The sample of 10,000 qualifier-tokens
(100 stimuli × 100 Ss) from each community was
analyzed at the Center for Comparative Psycholinguistics at the University of Illinois with computer
programs designed to determine for each qualifier-type
(a) its frequency of usage over all nouns, and (b) its
diversity of usage across the 100 nouns, and (c) its
independence of usage across the 100 nouns when
correlated with other qualifier-types. Measures (a) and
(b) can be combined in the single H-index of information theory statistics.
A list of qualifier-types, ordered in terms of H-rank
from most productive (highest frequency and diversity)
to least productive and culled for significantly correlated qualifiers, was returned to each language/culture
community. Common opposites were elicited for the
qualifiers, the highest ranking 50 pairs then being used
as the bipolar scales (e.g., good-bad, strong-weak, etc.)
against which the original 100 concepts were rated by a
different group of high-school-level males. These concept-on-scale judgments were then factor-analysed at
Illinois. The first factor in every case was clearly identifiable as Evaluation. Those qualifier antonym scales in
each language having loadings of ±.30 or greater on the
E factor were selected for analysis in the present study,
terms loading positively (e.g., good) being E+ and
negatively (e.g., bad) being E-. The H-rank, frequency
score and diversity score for each member of each
antonym pair (i.e., forgoodand for bad, for strong and
for weak, etc.) were determined separately from the
already available qualifier-elicitation data.
Results
T h e h y p o t h e s i s - - t h a t the E + member o f
each pair will be more frequent and diverse in
usage than the E - m e m b e r - - w a s tested by
comparing the H-ranks, frequencies and diversities o f the qualifiers constituting such pairs.
Table 1 summarizes these data. Listed for each
language/culture c o m m u n i t y are the numbers
of pairs displaying (a) higher H-rank, (b)
greater frequency, and (c) greater diversity for
3
either the E + or the E - members. (Languages
varied somewhat in the numbers o f scales
reaching the ~:.30 E fact0r-loading criterion,
and there were ties on some o f the measures for
some qualifier-pairs.) A l t h o u g h the H - r a n k is
an index o f b o t h frequency and diversity combined, it was felt that separate tests should be
made, just in case Pollyannaism should affect
frequency but not diversity, or vice versa. It
should be noted that in the cross-cultural project, the H, frequency and diversity measures
for individual qualifiers were always made
prior to the opposite elicitations and to the
factor analyses which determined qualifier
pairings and evaluative loadings.
The one-tail binomial probability o f the distributions a m o n g E + and E - occurring by
chance for each language and for the totals fcr
all languages were c o m p u t e d for the three
measures. The PoUyanna Hypothesis is clearly
supported. With a single exception (Finnish
on diversity), in every language E + words display higher average H-ranks, frequencies and
diversities than E - words. F o r seven languages
(French, Italian, Dutch, Mexican Spanish,
A f g h a n Farsi, Cantonese, and Turkish) the
differences are significant at the .05 level for all
measures, and for the totals the differences are
significant at the .001 level for all three
measures. The Pollyanna effect is weakest for
American English, Finnish, and Yugoslav
Serbo-Croatian. It can also be seen that the
effect is consistent across the three m e a s u r e s - the more diversely used E + words also tend to
be the more frequently used.
ANALYSIS I I : APPLYING NEGATIVE AFFIXES
CROss-LINGUISTICALLY
Negative affLxes are applied to an already
existing form in a language to reverse its meaning. I n English, such negations usually a p p e a r
as prefixes--indecent, immoral, dishonest,
unhappy, nonsense, and so forth. But since in
other languages negations m a y appear as
suffixes or infixes, as well as prefixes, we use
the more general term, affix. Negative affixes
4
BOUCHER AND OSGOOD
TABLE 1
NUMBERS OF QUALIFIERPAIRS FOR WHICH THE E + MEMBERIS HIGHER OR LOWER THAN THE E - MEMBERIN H-RANK,
FREQUENCY AND DIVERSITY
H-Rank
Language
Afghan Farsi
American English
Belgian Flemish
Cantonese Chinese
Finnish
French
Iranian Farsi
Italian
Mexican Spanish
Netherlands Dutch
Swedish
Turkish
Yugoslav Serbo-Croatian
Totals
Frequency
Diversity
E+
Higher
E+
Lower
E+
Higher
E+
Lower
E+
Higher
E+
Lower
18
21
18
28
18
24
18
23
24
18
19
23
17
6a
18
11
5"
15
12"
i0
11"
10"
10
10
7"
13
19
24
18
26
6"
15
11
5a
20
21
23
17
24
25
20
18
24
15
13"
13
11"
11"
10"
11
7"
14
19
21
18
28
23
23
19
32
25
20
19
24
16
6"
18
11
5"
20
13~
12
16"
12"
10~
11"
7"
15
269
138b
273
148b
287
15C
ap < .05.
~p < .001
could be used in either o f two ways: (a) A d d e d
to a positively ev al u a t e d w o r d to f o r m a negatively ev al u at ed opposite (e.g., happy-unhappy,
moral-immoral); (b) a d d e d to a negatively
e va l u at ed w o r d to f o r m a positively e v a lu a t e d
opposite (e.g., broken-unbroken, violent-nonviolent). T h a t the latter are relatively rare in
English can be d e m o n s t r a t e d by the reader
himself, by trying to think o f other instances.
T he P o l l y a n n a H y p o t h e s i s leads to the prediction that, for h u m a n languages generally,
negative affixes will be applied to E + w o r d s
(to m a k e E - opposites) m o r e f r e q u e n t l y t h a n
to E - wo r d s (to m a k e E + opposites). This is
because the hypothesis assumes priority o f E +
w or d s in a p p e a r a n c e in a language historically
as well as in the d e v e l o p m e n t o f language in
i n d i v i d u a l s - - a n d negative affixes must be
applied to already existent m e a n i n g f u l forms.
Method
The same qualifier data from the cross-cultural project on generality of semantic systems as used in Analysis I were used here as well. Identification of negative
affixes in each of the languages involved was obtained
from native speakers (when available on the Illinois
campus) or from linguists familiar with the languages.
The 50 final pairs of qualifier-opposites in each
language, defining its semantic differential scales, constituted the sample. Table 2 presents the frequencies of
applying negative affixes, when they appear at all
(English happy-sad would not contribute evidence;
happy-unhappy would), to either the E+ or the E members of opposite pairs of qualifiers. Cantonese
TABLE 2
FREQUENCIES OF APPLYING NEGATIVE AFFIXES TO THE
E + VS. E - MEMBERS OF QUALIFIER-OPPOSITE PAIRS
Language
Affix on E+ Affix on E Afghan Farsi
4
0
American English
4
1
Belgian Flemish
5
0~
French
5
0a
Iranian Farsi
8
1a
Italian
9
0a
Mexican Spanish
5
1
Netherlands Dutch
6
0"
Swedish
5
2
Turkish
2
0
Yugoslav Serbo-Croatian
10
0a
Totals
"p < .05 level.
~p < .001 level.
63
5b
THE POLLYANNA HYPOTHESIS
Chinese and Finnish had to be excluded, the former
qualifier data yielding no negative affixingand the
latter onlyone instance.
Results
As can be seen in Table 2, in every one of the
11 languages under investigation, negative
affixing is applied to the E+ member more
often than to the E - member, as predicted. For
six of the languages (Belgian Flemish, French,
Italian, Iranian Farsi, Netherlands Dutch, and
Yugoslav Serbo-Croatian), the differences are
significant at the .05 level by one-tail binomial
probability test. The totals are significantly
different at the .001 level. The Pollyanna
Hypothesis is again clearly supported.
The hypothesis leads to a correlated prediction: Where the negative affix is applied to the
E - word, its H-rank should be higher than the
E+ word, contrary to the general rule. In
other words, since negative affixes must be
applied to forms already available in a
speaker's vocabulary (and hence presumably
more frequent in the speaker's language
generally), application to E - words implies
that these words have priority of usage over
their E÷ opposites (if such exist at the time of
coinage). The derivation of nonviolent from
violent may well be a case in point. Checking
this prediction against the data in Table 2
(and eliminating all cases of tied H-ranks for
E+ and E - members of pairs), we find that in
every case where the negative affix is applied to
the E - member, this member has the higher
H-rank; where the negative affix is applied to
the E+ member, this member usually, but not
always, has the higher H-rank (42 out of 49
cases). Why there should be occasional cases
(seven in these data) that break the rule--the
E - term having the higher H-rank, yet the E+
term receiving the affix--will be considered in
the Discussion section.
ANALYSISIII: DEVELOPMENTOF E+ AND
E-- EVALUATION
In this analysis, we shift from comparisons
across languages and cultures to comparisons
5
across age levels within the American English
community. The Pollyanna Hypothesis leads
to the following expectations: (a) the E+ members of pairs should appear earlier in the
language of children than their E - opposites.
This could be due, in part, to the direct operation of Pollyannaism in humans, but it could
also be due to the greater availability of E+
qualifiers in the adult language.'(b) Frequency
and diversity of usage of E+ meml:ers of pairs
should be higher than their E - opposites for
all age levels, but the difference should decrease
with age. The decreasing E+/E- ratio would
be due to the relative increase in negative
evaluation with age. (c) Use of negative prefixes should be a late-occurring phenomenon,
and nearly all such affixing should be onto the
E+ member to make the E - opposite. The
prediction of late occurrence is consistent with
the delayed use of affixes generally in language
development.
Method
The data for this analysis were extracted from published articles by DiVesta (1964, 1965, 1966). The comparison with our cross-cultural analyses is very direct in
this case, since DiVesta applied almost identical procedures to those used cross-culturally by Osgood et al.
to comparisons across age levels--his groups being
7, 8, 9, 10, and 11 yr of age. The same 100 substantives
were used to elicit adjectival qualifiers from t00 boys of
each of these age levels.
Results
We may look first at the five highest-loading
E adjective-pairs (DiVesta, 1966) in relation to
their H-ranks at the different age levels (DiVesta, 1965). Table 3 presents these data. Note,
first, that in every case where one member has
an H-rank and the other does not, it is the E+
member that is ranked. This supports the first
prediction--that E+ words appear earlier than
their E - opposites in the course of language
development. Note, second, that in every case
where both members are ranked, the E+
member has the higher rank--supporting the
second prediction, that E+ members of pairs
will have higher frequencies and diversities
than their E - opposites.
6
BOUCHER AND OSGOOD
TABLE 3
H-RANKS OF PAIRED E + AND E - QUALIFIERSDEFINING THE FIVE HIGHEST LOADING E SCALESAS A FUNCTION OF
AGE LEVEL (DIVESTA, 1965, 1966)
Chronological Age
7
Scales
Good-bad
Pretty-ugly
Right-wrong
Sweet-sour
Funny-sad
8
9
E+
E-
E+
E- -
E+
E-
E+
E-
1
5
55
68
27
2
98
--61
1
6
29
-16
3
79
61
-38
1
9
59
63
19
3
48
68
-35
1
9
41
82
29
3
74
68
-60
l
14
72
-36
3
78
--57
.6_
o
¢.~
¢.o
~
0
S
.4_
m::
Z
°3-
.2-
.1
}
11
E-
Data collected from American Englishspeaking teen-age boys (ages 14-17) in the
cross-cultural project provide an "adult" criterion for the developmental data of DiVesta.
It will be recalled (cf. Analysis I) that a criterion of +.30 loading on the E Factor was
used to select a set of scales for comparative
analysis of the productivities of their E+ and
E - adjective-pairs. The same adjective-pairs
were analyzed here, comparing the mean Hscores (not ranks) for E+ and E - members at
each age level in the DiVesta (1965) data.
Figure 1 presents these data. The decreasing
scores for both E+ and E - members as a func-
.5_
10
E+
;
CHRONOLOGICAL
i
lo
fi
AGE
FIG. 1. M e a n H scores as a function of age in DiVesta
(1965) data for E + and E - words. Defining scales loading .30 or m o r e on E factor in Osgood et aL American
English cross-cultural data.
tion of age simply reflect an increasing qualifier vocabulary; that is, as the number of different qualifier-types available to the child
increases, the relative frequency of any particular type must decrease and hence similarly
the H-score. These data clearly fit the secondprediction--although the mean H-scores for
E+ members are larger than those for E members at all age levels, the magnitude of the
difference decreases with age.
The final step in this analysis was to check
the prediction relating to use of negative
affixes in the developmental data of DiVesta.
Inspection of the top 100 H-ranked qualifiers
at all age levels (DiVesta, 1965) revealed only
two instances of negative affixing, both applied
to E+ forms; such sparse usage precludes
quantitative analysis, but it is testimony to the
late appearance of negative affixing as expected. Inspecting the raw data for different
age levels (DiVesta, 1964), and noting first
those negatively prefixed qualifiers given more
than once to a noun in at least one age group,
we find only 10 types (different words) comprising only 64 tokens (occurrences), and this
by age 11. Keeping in mind that we have
approximately 10,000 qualifier-tokens (100
Ss × 100 noun-stimuli) for each age level, this
is a very small proportion. Noting, second, the
"idiosyncratic" responses--qualifiers given
only once to a given noun in a given age group
--we find that even here there are only four
tokens of four types at age 7 and only 114
tokens divided among 80 types at age 11. How-
THE POLLYANNA HYPOTHESIS
ever, the rate of increase in negative affixing
types is much more rapid between ages 9 and
11 than for total types, suggesting that during
this period of development the use of negative
affixes has been established and is generalizing.
Among the total 235 qualifier-tokens bearing a negative affix (qualifiers used either
idiosyncratically or more than once to nouns
in any age group), there are only 16 cases of
application to an E - term (the types being
unbroken, untied, undaunted, unforgettable,
unafraid, *unstale, independent, unharmable,
immortal, undying, undisturbed, unharmful and
unfailing); 25 of the token were unclear as to
evaluation (e.g., unusual, unchanging, unmarried); and all of the rerhainder (184 of the
total 235) involved affixing the negative to an
E+ term. These results are again consistent
with the cross-cultural data on the Pollyanna
Hypothesis.
DISCUSSION
The evidence reported here and elsewhere
(cf. Zajonc's review, 1968) indicates a strong
positive relation between productivity in
usage and favorableness in evaluation, but this
relation in itself does not say which is cause and
which is effect. The Pollyanna Hypothesis
asserts that evaluation is the cause of usage
frequency. The reverse causal relation is
asserted by Zajonc (1968)--that mere exposure to stimuli generates positive attitude or
evaluation. Thus frequency of usage creates
familiarity, and familiarity breeds not contempt but respect.
A careful consideration of Zajonc's paper
leads us to the following conclusions: First,
he-presents two quite different types of evidence. On the one hand, there is correlational
evidence for a positive relation between frequency of usage and evaluation of meaningful
words which are presumably near the asymptote of familiarity (e.g., possible vs. impossible).
These data are entirely consistent with the
results reported here. On the other hand, there
is experimental evidence for increasing evaluation as a function of frequency of exposure to
7
materials that are either meaningless (nonsense
words), unfamiliar (faces of unknown men), or
both ("Chinese" characters). There seems to
be no question but that unfamiliar stimuli-be they foreign foods or foreign faces or
Chinese characters--tend to be treated with
some combination of curiosity and disdain.
But the Pollyanna Hypothesis is concerned
with meaningful and familiar stimuli; the
7-year-old child is perfectly familiar with both
pretty and ugly, but he uses the former much
more frequently than the latter. Zajonc presents no evidence that stimuli which are both
meaningful and familiar can be made more
favorable in evaluation by mere exposure;
and, as a matter of fact, several studies he
reports which did meet these conditions
yielded negative results.
Second, the trouble with ~ the "frequencybreeds-value" hypothesis on theoretical
grounds is that it cuts words off from any
reasonable relation to their referents. High
frequency nouns like pain, thief, and fear
should end up with more favorable meanings
than low frequency nouns like ecstasy, respect,
and benevolence, and the very familiar bad
should be better than the relatively unfamiliar
reputable--none ofwhich makes any !ntuitive
sense. Another prediction contrary t o commonsense would be that as words come to be
used more and more frequently (e.g., communism, riots) they should gravitate away from
E - toward E+. Ex-Senator Joseph McCarthy
succeeded in making the Fifth Amendment
definitely less favorable in meaning (even for
college students) by repeatedly tying it to high-.
frequency Communist.
With regard to the use of the negative affix
in the cross-language study--where the rule is
application to the higher frequency E+ t e r m - why are there some cases where the affix is
applied to the E+ term even though the E term has a higher H-rank ? This is presumably
due to the fact that the use of negative words
(and negative affixing) begins to increase
rapidly around age 10, and some negative
evaluators thus may come to exceed their E+
8
BOUCHERAND OSGOOD
opposites in usage even though they appear
later in the language and hence receive the
negative affix. Also there is probably a
"semantic inhibition" working against assigning a negative affix (carrying a negative feeling
tone) to an already negatively evaluated word;
certainly *unsad and *unugly would not carry
the positive feelings of happy and beautiful
respectively !
It might be argued that, since the 100 nouns
used for qualifier elicitation in both the crosscultural work and Di-Vesta's developmental
studies were themselves biased 4 to 1 toward
E+, the frequencies and diversities of usage of
E + and E - qualifier opposites simply reflect
this bias. To the contrary, not only is the bias
in noun evaluations itself consistent with the
Pollyanna Hypothesis, but it is a linguistic
fact that any noun that can take an E+
qualifier could take an E - qualifier as well
(e.g., selfish love, bad luck, uneasy peace)---if
humans had an "Ebeneezer Scrooge" disposition. Girls could be ugly as easily as pretty,
fathers could be cruel as easily as kind and
friends could be sad as easily as happy,
linguistically speaking.
We are left with the question of why E+
words should tend to be used more frequently
by humans than E - words. Why do most people most of the time in most places around the
world talk about the good things in life ? Why
do they tend to see and report the good qualities of things ? The answer surely goes beyond
psycholinguistics per se and into the nature of
h u m a n social structures and the conditions
under which these structures can be maintained. It is hard to imagine human groups
whose members persistently look for and talk
about the ugly things in life and in their neighbors long remaining together. We leave further
elaboration to social philosophers.
REFERENCES
AMSTER, HARRIET,Evaluative judgment and recall in
incidental learning. J. verb. Learn. verb. Behav.,
1964, 3, 466-473.
ANISFELD,M., ANDLAMBERT,W. E. When are pleasant
words learned faster than unpleasant words ? J.
verb. Learn. verb. Behav., 1966, 5, 132-141.
BOUSFIELD, W. A., AND SEDGEWICK,C. H. W. An
analysis of sequences of restricted associative
responses. J. gen. Psycho/., 1944, 30, 149-165.
BOUSFIELD,W. A. An empirical study of the production
of affectively toned items. J. gen. Psychok, 1944,
30, 205-215.
D/VESTA, F. J. The distribution of modifiers used by
children in a word-association task. J. verb. Learn.
verb. Behav., 1964, 3, 421-427.
DIVESTA, F. J. Developmental patterns in the use of
modifiers as modes of conceptualization. Child
Develpm., 1965, 36, 186-213.
DIVESTA,F. J. A developmental study of the semantic
structures of children. J. verb. Learn. verb. Behav,
1966, 5, 249-259.
JOHNSON, R. C., FRINCKE, G., AND MARTIN, LEA.
Meaningfulness, frequency and affectivecharacter
of words as related to visual duration threshold.
Canad. J. Psycho/., 1961, 15, 199-204.
OSGOOD,C. E. Semantic differential technique in the
comparative study of cultures. Amer. Anthro.,
1964, 66, 171-200.
YAVUZ, H. S. The retention of incidentally learned
connotative responses. J. Psychok, 1963, 55, 409418.
ZA~ONC,R. B. Attitudinal effects of mere exposure. J.
person soc.Psycho/., Monogr. Suppl., 1968, 9, no. 2,
part 2, pp. 1-27.
(Received May 29, 1968)
Download