THE EFFECTS OF SOCIAL ENVIRONMENT ON
PAIRED-ASSOCIATE LEARNING AND
RETENTION IN NORMAL ELEMENTARY
SCHOOL CHILDREN*
CHUN-HSING CHANG
11
For many years
,
psychologists have showed considerable interest in the relationship between ability and verbal learning and retention. In general,
the research evidence is consistent with the common beHef: rate of acquisition of a verbal task is a positive function of initial level of abi1ity (6,
7,
8,
19,
21,
22). Data of this type suggest that verbal learning is a fertile field for research on learning and individual difference in inte1ligence.This generalization
,
however,
does not hold in the forgetting phenomenon. A1though an early study by Gi1lette (9) indicated that fast learners retained more than slow learners,
the data of recent studies (12,
26,
27,
28) indicated that the individual differen-ces presented in forgetting do not follow the same pattern as those in learning. That is,
there is no evidence that the rate of forgetting differs significantly between fast and slow learners.Group differences in learning and forgetting based on independent variables other than inte1ligence have not yet been investigated. The present study represents an attempt to investigate the effects of social environment on paired-associate learning and retention in two groups of normal elementary school children.
The major theoretical position from the experimental point of view with respect to forgetting places great emphasis upon interference as a critical factor (24
,
29). The interference is said to stem from the learning of other tasks fol1owing the learning of the task to be recal1ed (retroactive inhibition) and from tasks acquired prior to the learning of the task to be recal1ed (proactive inhibition). There has been a shift in the relative importance assigned to these two sources of interference. In earlier years,
retroactive inhibition was assumed to be the major source of interference in producing forgetting. Recently,
however,
Underwood (29),
Underwood and Postman (32) and Postman (24) have assigned the major role to proactive inhibition. The reason for this switch in emphasis comes mainly from the concept of spontaneous recovery of old habit being temporarily suppressed during the learning of a new habit. Thus spontaneous recovery of prior learning producing proactive inhibition is assumed to be an important variable in long-term memory. This assumption has been supported by research evidences (2,
1,
*
This study was supported by a research grant from the National Council of Science Development,
1971-1972.25, 15). This concept has been widely accepted (14, p. 501) and been extended by Underwood and Postman (32) and McGovern (20) in terms of ordinary language habits (letter-sequence interference and unit-sequence interference) and multidirection associations (S-R and R-S
,
etc.).This development has brought about a fundamental change in the interference theory and has stimulated a new direction of research. Several hypotheses have been developed based on the concept. Klausmeier et aI. (16)
,
for example,
hypothesized: (1)high IQ children would learn rapidly and forget rapidly whereas low IQ children would learn slowly and forget slowly; and (2) retention would be the same among children of low
,
average,
and high intelIigence when the tasks are learned to the same criterion. Underwood (31) hypothesized: (1) very young children would show less rapid forgetting of a verbal task than colIege students because of the fewer conflicting associations in children's repertoire to cause proactive inhibition; and (2) if an imbecile were given enough trials on a verbal task to reach the same degree as that attained by a genius, the genius would show more rapid forgetting because he has learned more things and wilI subsequentIy learn more things that wiII interfere.These kinds of hypotheses have been tested by a number of studies in the last decade
,
but the resuIts are conflicting. Some findings indicated no difference among groups with different IQ levels either on original learning or on retention (5, 13, 18,). Some studies found differences on rate of learning, but not on retention (3 ，時， 17). StiII others indicated no differences among abiIity levels on learning, but significant differences on retention (23, 33).In general, the previous studies have three similarities: (1) verbal tasks were used as learning material and the single paired-associate learning technique was employed; (2) measures of long-term memory were used as criterion of retention; and (3) two (or more) IQ groups
,
principalIy normal and mentalIy reatarded children,
were used as subjects. Based on the original hypotheses proposed by Klausmeier (16) and Underwood (31),
and according to the current version of the interference theory,
it seems logical to assume that the degree of proactive and retroactive interference is greater for normal children than for retardates because of their more diverse verbal experience. From the metIf other variables rather than verbal experience were equal
,
the less inteIligent children might forget less rapidly because of less interference. Unfortunately, the experiments comparing normals and retardates on learning and retention permit no degree of experimental control over the related variables. Underwood (30,
p. 123),
for example, suggested that if we separate subjects on the basis of a particular charac-teristic (e.g.,
intelIigence) and then test them on an aspect of bèhavior not used in theEffects of Social Environment on Learnig and Retention 13
original separation
,
the groups should be equated on al1 other variables other than those used to effect separation. More recently,
Goulet (10) pointed out that mentally retar-ded subjects may have an additional variable of structural memory deficiency. If this is the case, the course of forgetting over time in the normals and in the retardates would be different. Verbal forgetting in normal children might be attributable to only two variables,
namely,
proactive and retroactive interference. In the retardate group,
however
,
the course of forgetting must be considered to be a joint function of three variables, proactive and retroactive interferences and the structural deficiency.Because of the methodological weakness, the results of the studies comparing normals and retardates are subject to difficulty in interpretation. This led Baumeister (3) to conclude
“
To understand the behavior of retardates one must study the behavior of retardates. The study of normal behavior is quite irrelevant to this purpose." To the same effect, to understand the behavior of normal one must study the behavior of normals. For this reason,
the present study instead of comparing normals and retardates in learning and retention,
two normal groups drawn from different social environments employed as subjects. According to the current version of the interference theory,
it is reasonable to assume that individuals living in a complex and enriched social environment would have more verbal stimulations than that of living in a simple and deprived social environment. These differences may influence children's verbal experience in producing different effects upon their learning and retention. The present study was designed to investigate the effects different verbal environment on children'spaired-associate learning and retention.
In viewing of our language habits in terms of the concepts of unit-sequence interference (32) and multi-direction associations (20), they seem to be more relevant to Chinese than to English. The basic expressions in the Chinese language are not single words (characters) but terms
,
which are combinations of two or more single words. Chinese children have to learn the meanings of the single words as wel1 as the derived new meanings of the terms. In considering the variety of Chinese lan-guage,
two features,
at least,
of the basic unit of Chinese language should be identifiedretention of learned material based on the number of correct responses obtained after different time intervals.
METHOD
Subjects Fifty normal children (25 boys and 25 girls, IQ range 90 to 115), who were enrolled in elementary public school fourth grade classes in an urban area (Taipei City) were selected as subjects for the urban normal group (UNG). In addition to the variable of intelligence determined by a standardized non-verbal intelligence test
,
the UNG subjects were selected on the basis of the following criteria. All of the chi1dren had attended kindergartens at least one year. Their parents had at least high school educations. They had lived in the same environment since birth. Most of their families provided reading materials above and beyond school textbooks.
Another fifty normal children (25 boys and 25 girls
,
IQ range 90 to 115),
who were enrolled in elementary public school fourth grade classes in a rural area were selected as subjects for the rural normal group (RNG). The selection of this group was based on the following criteria. None of the children had preschool education. Their parents' education did not extend elementary school. They had lived in the same environment since birth. Most of their homes did not provide reading materials other than school textkooks.MateriaIs The learning materials for this experiment consisted of 12 paired-associate verbal tasks (unrelated pairs of Chinese characters). Each of the paired-associate words was printed in black ink on a 4" x 6" white card. These words were selected on the basis of two criteria: (1) all words had to be known by the subjects; the subjects' task was to learn the associations of the paired words
,
not the separate words themselves; and (2) according to the textbooks previously used by the subjects and the Chinese language used daily,
none of the paired-associates could be a meaningful term; that is,
the learning task had to be new to the subjects and the pairedassociate words had to be unrelated.Procedure Since one of the purposes of the study was to investigate the relation-ship between subjects' verbal experiences and their learning environments
,
it was necessary to devise a technique for measuring verbal complexity. Therefore,
afree-recall test was given to the subjects before experiment. A list of 15 Chineses words drawn from textbooks used previously by the subjects was given to the subjects
,
and they were required to use them as cues for making word-combination terms. In other words,
the subjects were asked to recall freely the words associEffects of Social Environrnent on Learnig and Retention 15
number of the terms an individual recalled represented his verbal complexity
,
which was used as an index of the level of language development.The experiment was conducted in a small room in the school. The experimenter
(E) worked with subjects (5) individually. The subjects was seated at a small table facing E. After E recorded the name and birth date
,
the following instructions were g1Ven:We are going to play a game with these cards. 1 want you to read each of the two words on the card (E points to the card). Try to remember that the two words go together, because later 1 am going to show you just one word (E shows him the stimulus word on the back of the-c.ird) and ask you to tell me which word goes with it. Do you know how we play? (if 5 indicates understanding) OK
,
let's begin now.E then manually presented each card to 5 at a rate of approximately five seconds per card
,
with 5 saying the words aloud. After one trial,
exposing all 12 cards to 5,
E shuffled the cards and said:Now 1 am going to show you only one word on the opposite side of the card. Would you please tell me which word goes with it?
The cards were presented at a rate of approximately 5 seconds per card
,
with 15-second intertrial interval. At the end of each trial the cards were shuffled during the 15-second interval in order to control for possible serial effect. If 5 made an incorrect response to the stimulus word,
E turned the card and 5 verbally corrected his response to the stimulus. Following a correct response, E responded with“
good,"“
that's fine," or other expressions of approval and rE'frained from the card.The material was learned by the adjusted-learning method originated by Woodworth
(3日 and adapted and refined by Gillette (9). By this method the cards in the deck were dropped out after 5 correctly responses. Again
,
when the number of cards in the deck reached zero,
the deck was reassembed a third time and readministered to 5. Thus,
each subject made three correct responses to each of the stimulus words.Subsequent to the attainment of the criterion
,
retention was measured by presenting 5 with only the stimulus words and recording correct responses. The time interval between termination of learning and the first retention test was one hour. When retention was measured,
incorrect responses were not corrected by turning cards These procedures were repeated one day (土 2 hours),
and 15 days (土 4 hours) after initial training session.RESULTS
Original Learning The means and standard deviations of trials to criterion as well as the mean number of trials needed to reach the point of error1ess response at different time of learning practice for the two groups are presented in Table 1. Since the main purpose of the study was to investigate the effect of social learning
environ-ments on pairedassociate verbal learning
,
the difference in the measures of learning of UNG and RNG should be tested first. Test of significance (t-test) applied to these data indicated that the difference between these two groups was not statistically sig-nificant.To analyze the effect of interaction of social groups and the measures of learning
,
a 2 x 3 complete factorial analysis of variance with repeated on one factor (34) was used. The groups (UNG and RNG) and the three times of learning practice to reach the points of errorless response served as the independent variable
,
and mean numbers of trials needed for reaching the points served as a dependent variable. The resu1ts are shown in Table 2. The data indicated that the main effect of subject cJassification (UNG and RNG) was not significant. There were no effects attributable to the interaction between the subject cJassification and the mean numbers of trial needed to reach the points of errorless response for each time of learning practice. The data show cJearly that both of the groups improved significantly (F=173.90,
p<.Ol) through the three time of learning practice. The first hypothesis of the present study,
there-fore,
was not rejected.To provide more information concerning paired-associate verbal Jearning as reJated to environmental and sex factors
,
a 2 x 2 complete factorial analysis of variance was used to further analyze the data. The resu1ts are presented in Table 3,
and the summary of the analysis of variance is shown in Table 4. As indicated in Table 4,
the differences between boys and girls and between UNG and RNG were not significant. The effects of interaction of these two variables were also not significant.
Verbal Complexity The means and standard deviations of measures of subjects' verbal complexity two groups obtained by the method of free recall are presented in Table 5. Tests of significance (t一test) applied to these data indicated that the RNG Ss were significantly inferior to UNG (t=4.51, p<.Ol). This resu1t indicated that the verbal development was a function of social learning environment.
Table 1 Means and Standard Deviations of Trials to the Point of Errorless Response and to Criterion
I First Time I Second Time I Third Time I Total
Reaching I Reaching Reaching ! Number
the Point the Point ! the Point I of Trials
of Errorless I of Errorless of Errorless ; to
Response
I
Response i Response ) CriterionGroup N
I心也~IMean
I-SD;
Mean SD!Me;三n
I SDUNG 回 7.78 I 2.98 4 布 1 啞 3 曲( 1.38 16.02 5.44
17
Effects of Social Environment on Learnig and Retention
Summary of the Analysis of Variance of Trials Needed for Reaching the Point of Error less Response
Table 2 172.90** 1.58 .56 F 9 臼 戶 hd
-9UH 一 21.34 13.54 438.23 1.40 MS 1373.33 876.45 2.80 494.08 2723.59 df SS 99 1348.26 21.34 1326.92 2 2 196 299 200 98 1 Source Between Subjects Between Groups Subjects within GroupsWithin Subjects Between Times Interaction (Groups x Times) Times x Subjects within Groups Total **p<O.Ol Girls 15.88 (n=25) 19.72 (n=25) Means of Trials to Criterion of Paired-associate Learning as Related to Environmental and Sex Factors
Table 3 Boys 16.16 (n=25) 15.60 (n=25) Group UNG RNG
Summary Analysis of Variance of Trials to Criterion of Paired-associate Learning as Related to Environmental and Sex Factors Table 4 F 1. 74 3.14 2.37 MS 92.16 67.24 38.51 121 SS 67.24 92.16 d
一一
1 1 1 Source Between(Boys and Girls) Between
(UNG and RNG) Interaction
(Sex Groups x Social Groups)
Within Total
Group
UNG RNG
Table 5 Means and Standard Deviations of Measures of Verbal Complexity N 50 50 Mean 125.74 106.92 SD 19.05 22.54
Retention The means and standard deviations of the correct responses of different groups on the tests of retention of learned material at different time intervals are presented in Table 6. A 2 x 3 complete factorial analysis of variance with repeated measures on one factor was used to analyze the retention data
,
and the results are presented in Table 7. These tables indicated,
as expected,
that the mean numbers of correct responses measured at different intervals were significantly different (F=222.鈕，p<.Ol). That is
,
the recall of the learned material was significatly reduced as time passed. Inspection of Table 7 indicated that differences between groups were also significant (F=16.妞， p<.01). In order to determine which comparisons among the three means were significant,
t一test was applied,
and the resu1ts are shown Table 6. The results indicated that the retention scores of RNG were significantly superior to UNG at all three tests of retention obtained at different time intervals after original learning. The effect of interaction of groups and intervals was also significant.Based on the results analyzed the second hypothesis that there is no significant difference between groups in the measures of retention was
,
therefore,
rejected.The measures of retention as related to environmental and sex factors were analyzed. The difference between boys and girls was not significant.
Table 6 Means and Standard Deviations of the Correct Responses of two Groups on Tests of Retention of Learned Material
1 hour Group N Mean SD UNG 50 8.14 1.86 RNG 50 9.72 2.18 t 料p<O.01 Retention Interval 1 day Mean SD 6.90 8.02 2.93料 1.73 2.08 15 days Mean 4.94 6.78 4.34** SD 2.09 2.15
Effects of Social Environment on Learnig and Retention 19
Table 7 5ummary of the Analysis of Variance of 5cores of Two Groups on Tests of Retention of Learned Material
50urce F
Between Subjects
Between Groups 5ubjects within Groups
Within Subjects Between Intervals Interaction (Groups x Intervals) Intervals x 5ubjects within Groups Total 而p三三。而I 可乏正的 df 55 M5 1166.92 171. 77 995.15 686.00 471.53 n曰， inmuhuqd nynyhu qh 171. 77 10.15 16.92** 235.77 222.42料
斗
3.13* 2 6.64 3.32 196 207.83 1.06 1852.92DISCUSSION
According to the interference theory and based on the facts that the children from culturally deprived environment were retarded in verbal development
,
it would be reasonable to assume that the subjects in RNG would have fewer old associations with the familiar stimulus words to b 三 extinguished during the acquisitions of new associa-tions. This might reduce the proactive inhibition produced by previous associations and consequent1y result in faster learning. In other words,
the group which had lower verbal complexity would need fewer trials to criterion. The results of the present study,
however,
did not support this assumption. Contrary to expectations,
when two groups were combined and mixed up without regard for social classification,
a negative correlation (r= -
.23,
p< .05) was found between the measures of learning and the measures of verbal complexity. That is,
the subjects who had more complex language habits learned faster than those who had simpler language habits. Based on the con-cept of interference,
this rather tangled problem could not be readily answered. This phenomena might be explained in terms of extraexperimental transfer proposed byGoulet (11).
Fundamental旬， there are two paired-associate transfer paradigms: A-B, A-C; and A-B
,
C-B. The designation implies that subjects learn two consecutive paired-associate lists. The A-B,
A-C paradigm,
involving identical stimuli and unrelated responses on the two lists,
typically results in negative transfer. This negative transfer is attributed to interference where the first-list association (A-B) inhibits the new associations of the second-list (A-C). In the A-B,
C-B paradigm,
involving unrelated stimuli and identical or similar responses,
a positive transfer is expected. This positive transferstem from the transfer of response 1earning from the first-list to the second-1ist. 1n the two-list experiment these types of transfer can readily be identified. Now could we use these concepts to interpret the peculiar phenomena of the obtained data?
1n this experiment
,
instead of a two-list design,
a single paired一associate task was employed. And particulary,
both the stimulus and response items were meaningful Chinese words known previously by the subjects. These words,
a1though unrelated in the designated pairs,
were apparently associated with other words in the subjects natural language habits (word-combinationterms). If we take the learned single paired-associate task as A-C, and assume the stimulus item“
A" had a number of preexperimen-tal learned associations such as A-B, A-D, A-E, . . . .; the higher the verbal complexity, the more preexperimental associations. If this is the case, extraexperimental transfer would be negative (since the conceptualized paradigm is A-B, A-C).However
,
the single paired-associate A-C should be considered as a mu1tireaction association (20). Since both“
A" (the stimulus item) and“
C" (the response item) are meaningful words, they would not only have their own preexperimental associations, but also be associated in different ways. The following paradigm could be conceptua一lized:
B-
C
(Preexperimental association)
A-C
(The single paired-associate)
If this is the case
,
the extraexperimental transfer would resu1t in a positive effect.Furthermore, the preexperimental associations could be in various forms such as B-C, D-C, E-C, . . . ., with identical responses and unrelated stimuli. It follows from this paradigm that the higher verbal complexity and the more preexperimental associations
,
the better facilitation learning condition would be.
With these conceptual analyses above
,
the extraexperimental transfer might be either positive or negative in nature. The findings in the present study that the higher verbal complexity,
the faster the learning to criterion,
might be attributable to the positive extraexperimental transfer.SUMMARY
Based on the current version of the interference theory
,
the effects of children's verbal learning environment on paired-associate learning and retention was examined. Two groups, UNG (n=50) and RNG (n=50), with equal 1Q levels were drawn, based on defined selected criteria, from two social environments. The learning material was 24 unrelated Chinese characters, and the single paired-association technique was used. The results indicated: (1) The difference of measures of learning to criterion between two groups was not significant. (2) The difference of measures of verbal complexity measured by free recall revealed that the RNG Ss were significantly inferior to Ss in UNG. (3) RNG Ss retained significantly superior to Ss in λJNG. (4) No significant difference between boys and gir1s was found either in learning or in retention. (5) The interference theory was partially supported by the findings of the present study.Effects of Social Environment on Learnig and Retention 21
REFERENCES
( 1) Adams, S. Temporal Changes in the Strength of Competing Verbal Associations.
Unpublished Doctoral dissertation
,
University of California,
Berkley,
1961.( 2) Barnes,
J.
M. and Underwood, B.J.
“
Fate" of first-list association in transfer theory. Journal of Experimental Psychology,
1959,
58,
97-105.( 3) Baumeister, A. A. Problems in comparative studies of mental retardates and normals. American Journal of Mental Deficiency
,
1967,
71,
869-875.( 4) Briggs, G. E. Retroactive inhibition as a function of the degree of original and interpolated learning. Journal of Experimental Psychology
,
1957,
53,
60-67.( 5) Cantor
,
G. N.,
and Ryan,
T.J.
Retention of verbal paired-associates in normals and retardates. American Journal of Mental Deficien旬， 1962 ，駒， 861-863.( 6) Cieutat
,
V.J.
,
Stockwell,
F. E.,
and Noble,
C. E. The interaction of ability and amount of practice with stimulus and response meaningfulness in paired associate learning. Journal of Experimental Psychology,
1958 ， 時， 193一2一2.( 7) Duncanson,
J.
P. Learning and measured abilities. Journal of Educational Psychology,
1966， 肘， 220-229.( 8) Games, P. A. A factorial analysis of verbal learning tasks. Journal of Experimen-tal Psychology
,
1962,
63,
1一1 1.( 9) Gi11ette, A. L. Learning and retention: a comparson of three experimental procedures. Archives of Psychology
,
1963,
28,
No. 198.(1的 Goulet ， L. R. Verbal learning and memory research with retardates: an attempt to assess development trends. In Illis, N. R. CEd.), International Review of Research in Mental Retardation
,
Vol. 3,
New York: Academic Press,
1968,
67一173.(11) Goulet, L. R. Verbal learning in children: Implication for developmental research. Psychological Bulletin
,
1968,
69,
359一376.(12) Grogory
,
S. C.Jr.,
and Bunch,
M. E. The relative retentive abilities of fast and slow learners. Journal of General Psychology,
1959， 駒， 173-181.(13) Heber
,
R. F.,
Prehm,
H.,
Nardi,
G.,
and Simpson,
N. Learning and retention of retarded and normal children on a paired-associate task. Paper read at the annual meeting,
American Association of Mental Deficiency,
New York,
May 1962.(14) Hilgard
,
E. R.,
and Bower,
G. H. Theories of Learning. New York:Appleton-Century-.Crofts, 1966.
(15) Keepel
,
G. Retroactive and proactive inhibition. In Dixon,
T. R.,
and Horton,
D. L. CEds.) Verbal Behavior and General Behavior Theory. Englewood Cliffts
,
N.J.:
Prentice-Hall, 1968.(17) Lance
,
W. D. Effects of meaningfulness and overlearning on retention in normal and retarded adolescents. American Journal of Mental Deficiency,
1965,
70,
270-275.
(18) Lott
,
B. E. Paired-associate learning,
generalization,
and retentions as function of intelligence. American Journal of Mental Deficiency,
1958， 缸， 481-489.(19) Mandler ，丘， and Huttenlocher
,
J.
The relationship between associative frequency,
associative ability
,
and paired-associate learning. American Journal of Psychology,
1956
,
69,
424-428.(20) McGovern
,
J.
B. Extinction of associations in four transfer paradigms. PsychologicalMonogra，ρhs， 1964
,
78,
whole No. 593.(21) Noble
,
C. E. Abi1ity vs practice in paired-associate learning. Journal of Psycho-logy
,
1959,
47,
331-335.(22) Noble
,
C. E. Verbal learning and individual differences. In Cofer,
C. N. (Ed.)Verbal Learning and Verbal Behavior
,
New York: McGraw-Hill,
1961,
132一 146.(23) O'Conner
,
N.,
and Hermelin,
Beate. Recall in normals and subnormals of like mental age. Journal of Abnormal and Social Psych心logy， 1963， 前， 81-84.(24) Postman
,
L. Extra-experimental interference and retention of words. ]i仰rnal of Experimental Psychology,
1961 ， 缸， 97-110.(25) Postman
,
L. Does interference theory predict too much forgetting? ]i仰:rnalofVerbal Learning and Verbal Behavior
,
19白， 2,
40-48.(26) Shuell
,
T.J.
,
and Keepel,
G. Learning ability and retention. Journal of Educationa. Psychology,
1970， 缸， 59-65.(27) Stroud
,
J. 且， and Schoer,
L. A. Individual differences in memory. Journal ofEducati，仰al Psychology
,
1959， 間， 285一292.(28) Underwood
,
B.J.
Speed of learning and amount of retained: a consideration of methodology. Psychological Bulletin,
1954， 日， 276-282.(29) Underwood
,
B.J.
Interference and forgetting. Psychological Revie仰， 1957,
64,
的-60.(30) Underwood
,
B.J.
Psychological Research. New York: Appleton-Century一Crofts，1957.
(31) Underwood
,
B.J.
Degree of learning and the measurement of forgetting. Journal ofVerbal Learning and Verbal Behavior,
1964,
3,
112-129.(32) Underwood
,
B.J.
,
and Postman,
L. Extra-experimental sources of interference in forgetting. Psychological Review,
1960， 前， 73-95.(33) Vergason
,
G. A. Retention on retarded and normal subjects as a function of amount of original learning. American Journal of Mental Dejiciency,
1964， 駒，623--629.
McGraw-Effects of Social Environment on Learnig and Retention
