28
COGNITIVE PSYCHOLOGY 10, 1-28 (1978) Memories of Morocco: The Influence of Age, Schooling, and Environment on Memory DANIEL A. WAGNER University of Pennsylvania To what extent are models of memory general, in that they may be applied to children or to other cultural groups? In an attempt to answer this ques- tion, two experiments were undertaken in Morocco to investigate various cultural and experiential antecedents to memory development. A total of 384 children and young adults, ranging in age from 6 to 22 years, were tested in a design that contrasted schooled and nonschooled children in urban and rural environ- ments. Three additional groups of subjects-Koranic students, Moroccan rug sellers, and University of Michigan students-were also studied because it was hypothesized that each might have particular “culture-specific” memory skills as a function of previous experience. A serial short-term recall task was used in Experiment I. Results showed that the recency effect or short-term store was generally invariant with age or experience. Control processes appeared to be a function of age, but only when coupled with schooling, and, to a lesser extent, urban environment. In Experiment II, a continuous recognition memory task was given with black and white photographs of Oriental rugs as stimuli. Forgetting rates were generally invariant with age and experience, while the acquisition parameter seemed to vary as a function of specific cultural experiences. Data from the three additional groups were useful in supporting the hypothesis of culture-specific memory skills. From Experiments I and II, and previous research, it was hypothesized that structural features of memory (e.g., short-term store and invariant forgetting rates) may be universal, while control processes or mnemonics in memory are probably culture-specific, or a function of a variety of experiential and cultural factors that surround the growing child. This research was supported, in part, by a Foreign Area Fellowship from the Social Science Research Council, and by a traineeship from the National Institute of Child Health and Human Development. This research was also described in a doctoral dissertation submitted in partial fulfillment of the requirements for the degree of Ph.D. in Psychology at the University of Michigan (1976). The author is indebted to several ministries ofthe Moroccan government, as well as to many individual Moroccans who helped in innumerable ways to make this research possible. Among these were: M. Akoujane, A. Benabderslam, S. Katim, and especially M. Mouhcine. The author would also like to thank J. Baron, J. Hagen, R. Kail, G. Olson, and H. Stevenson and the reviewers (T. Landauer and S. Scribner) of Cognitive Psychology for their very helpful comments on earlier drafts of this paper. Reprints of this paper may be obtained from the author at: Department of Human Learning and Development, Graduate School of Education, University of Pennsylvania, Philadelphia, PA 19174. 1 OOlO-0285/78/0101-0001$05.00/0 Copyright 0 1978 by Academic Press, Inc. All rights of reproduction in any form reserved.

Memories of Morocco: The influence of age, schooling, and environment on memory

Embed Size (px)

Citation preview

COGNITIVE PSYCHOLOGY 10, 1-28 (1978)

Memories of Morocco: The Influence of Age, Schooling, and Environment on Memory

DANIEL A. WAGNER

University of Pennsylvania

To what extent are models of memory general, in that they may be applied to children or to other cultural groups? In an attempt to answer this ques- tion, two experiments were undertaken in Morocco to investigate various cultural and experiential antecedents to memory development. A total of 384 children and young adults, ranging in age from 6 to 22 years, were tested in a design that contrasted schooled and nonschooled children in urban and rural environ- ments. Three additional groups of subjects-Koranic students, Moroccan rug sellers, and University of Michigan students-were also studied because it was hypothesized that each might have particular “culture-specific” memory skills as a function of previous experience.

A serial short-term recall task was used in Experiment I. Results showed that the recency effect or short-term store was generally invariant with age or experience. Control processes appeared to be a function of age, but only when coupled with schooling, and, to a lesser extent, urban environment. In Experiment II, a continuous recognition memory task was given with black and white photographs of Oriental rugs as stimuli. Forgetting rates were generally invariant with age and experience, while the acquisition parameter seemed to vary as a function of specific cultural experiences. Data from the three additional groups were useful in supporting the hypothesis of culture-specific memory skills.

From Experiments I and II, and previous research, it was hypothesized that structural features of memory (e.g., short-term store and invariant forgetting rates) may be universal, while control processes or mnemonics in memory are probably culture-specific, or a function of a variety of experiential and cultural factors that surround the growing child.

This research was supported, in part, by a Foreign Area Fellowship from the Social Science Research Council, and by a traineeship from the National Institute of Child Health and Human Development. This research was also described in a doctoral dissertation submitted in partial fulfillment of the requirements for the degree of Ph.D. in Psychology at the University of Michigan (1976).

The author is indebted to several ministries ofthe Moroccan government, as well as to many individual Moroccans who helped in innumerable ways to make this research possible. Among these were: M. Akoujane, A. Benabderslam, S. Katim, and especially M. Mouhcine. The author would also like to thank J. Baron, J. Hagen, R. Kail, G. Olson, and H. Stevenson and the reviewers (T. Landauer and S. Scribner) of Cognitive Psychology for their very helpful comments on earlier drafts of this paper.

Reprints of this paper may be obtained from the author at: Department of Human Learning and Development, Graduate School of Education, University of Pennsylvania, Philadelphia, PA 19174.

1 OOlO-0285/78/0101-0001$05.00/0 Copyright 0 1978 by Academic Press, Inc. All rights of reproduction in any form reserved.

2 DANIEL A. WAGNER

Many contemporary models of memory distinguish between two major features of human memory: structure and controlprocesses (e.g., Atkinson & Shiffrin, 1968). Structure is said to be composed of a sensory store, short-term store, and long-term store. The sensory and short-term stores are said to have relatively fixed capacities, while the capacity of long-term store is relatively unlimited. Information is assumed to be forgotten at fixed, but different rates from each store. Thus, forgetting rates are also considered to be structural features of memory. Information is guided through the memory structure by control processes. These control processes are sometimes, but not always, under the conscious control of the individual. Strategies such as verbal rehearsal, clustering, and categorization are considered to be typical control processes employed by individuals to aid in their remembering.

These models of memory have been derived from a composite of research with adults. However, similar distinctions between structure and control processes have been made in discussions of memory development with children. Evidence for the early emergence and relatively unchanging structure of memory with age comes from a variety of studies. The capacity of sensory or “iconic” (after Neisser, 1967) store for rapidly presented visual material was found to be unchanging in children from as early as six years of age (Morrison, Holmes, & Haith, 1974). In serial memory tasks, the recency effect-an indicator of short-term store’-has been found to be relatively invariant with age from early childhood (Hagen, 1971; Wagner, 1974). While the capacity of short-term store had been thought to increase with age, it may be that such improvements were due to usage in older subjects of different encodings or “chunking” of information units (Chi, 1976; Olson, 1973; Simon, 1974). Furthermore, rates of forgetting in pictorial recognition tasks seem to be invariant from young childhood to adulthood (Brown, 1975), and through old age (Craik, 1971; Wickelgren, 1975). Such studies seem to support the hypothesis that the structure of memory changes little from early childhood.

1 Despite a large research effort devoted to its study, explanation of the recency effect remains a controversial topic. In serial short-term memory tasks, some investigators (e.g., Hagen, 1971; Hagen et al., 1970) have suggested that the recency effect is a function ofechoic memory (after Neisser, 1967). While such usage may be appropriate for those subjects who provide spontaneous verbal labels to the pictures presented, echoic store may be an inappropriate term for those who do not spontaneously label. In contrast, Baddeley and his associates (Baddeley, 1976; Baddeley & Hitch, 1977) have suggested that the recency effect may be a low-level retrieval strategy, controlled by a central processor in primary or “working” memory. In order to avoid controversy over the choice of words such as primary memory, working memory, and short-term store, it would be tempting to use the term “recency mechanism” to stand for that aspect of memory that continually produces recency effects in numerous experiments. Since the present discussion is based primarily on the model of Atkinson and Shiffrin (1968), their expression “short-term store” will be used here for consistency.

MEMORIES OF MOROCCO 3

The development of control processes has been the focus of a large body of research with children. In contrast to structural features of memory, control processes seem to develop much more slowly in the individual. The progressive development of mnemonics such as verbal rehearsal, clustering, categorization, semantic elaboration, and meta-memorial skills is well documented (cf. Brown, 1975; Cole, Frankel, & Sharp, 1971; Flavell, 1970; Flavell& Wellmen, 1977; Hagen, Jongeward, & Kail, 1975). The development of these control processes has been found to begin as early as three years of age (Wellman, Ritter, & Flavell, 1975), and appear in stable or spontaneous form by the age of 13 years.

In sum, studies of memory development in children suggest that age-related increases in performance are mainly a function of developmen- tal changes in control processes, rather than in structural features which appear relatively early in childhood.

The present study provides a potential test for the developmental model of memory outlined above. If memory structure is more or less invariant with age, then different experiences in childhood should have little effect on the emergence of such structure. However, if control processes develop more slowly through early and middle childhood, then we may be able to determine the extent to which varying degrees of early experience can account for such development. That is, if control processes are a function of certain childhood experiences, then children with greatly different backgrounds might exhibit control processes that differ to some measurable degree. In order to determine the specific experiences that might underlie such development, a cross-cultural* design was used to study populations of children that vary greatly in degree of early experience or stimulation.

The cross-cultural approach in psychology is an important research strategy for a least two reasons: First, it allows the investigator to examine the generality of psychological statements, based on data collected from restricted subject populations, to other populations; and second, it allows the investigator to study previously confounded or noncontrastable variables-such as years of age and years of schooling-whose ranges are constrained by the cultural and environmental limitations of Western society. While there is a long and varied history of comparative research on human memory (e.g., Bartlett, 1932; Levy-Bruhl, 1910; Porteus, 1937), in recent research, the cross-cultural strategy has been primarily used to

2 The term “cross-cultural” may seem to be out of place here, as the experiments were conducted primarily within a single country, Morocco. In current usage, and in the present context, a cross-cultural design may also be one that is able to study and tease apart cultural or environmental variables within one country. There are clearly many advantages with this design over earlier cross-cultural research which might have compared one group of American college students with one group of nonschooled Moroccans-leaving a multitude of uncontrolled cross-cultural variables.

4 DANIEL A. WAGNER

study the effects of different early environments on subsequent cognitive development. Implicit in such cross-cultural studies is the need to contrast the kind of antecedent experiences that are likely to produce measurable differences in behavior, or, if such differences do not occur, to support the notion of psychological universals.

One environmental variable that has attracted considerable inter- est is that of formal schooling. The powerful effects of schooling have been found in a number of studies, perhaps best exemplified by Greenfield’s study of Piagetian conservation among Senegalese children (Bruner, Olver, & Greenfield, 1966). She found that urban and rural schooled children were more likely to “conserve” than nonschooled rural children. The positive effects of formal schooling on memory have also been studied. For example, Wagner (1974) showed that by age 14, urban in-school children of Yucatan were performing better than rural out-of-school children on a serial short-term memory task. While few differences were found in the recency effect by age or by schooling, large differences emerged in the primacy portion of the serial position curves. It was hypothesized that the high performance of urban schooled children was a function of mnemonic strategies such as verbal rehearsal. In a series of studies, Cole and his associates (Cole, Gay, Glick, & Sharp, 1971) found that schooled Liberians showed enhanced memory abilities over nonschooled Liberians. Among the important differences was the spontaneous use of clustering strategies in free recall by schooled children. Another study in Liberia showed that the degree of urbanization for certain illiterates was related to their clustering strategies and produced increased performance on a free recall task (Scribner, 1974). In a more recent study in Mexico, free recall performance was found to be a function of number of years of schooling, but independent of standard SES factors (Sharp, Cole, & Lave, Note 2).

The present study was designed to consider the effects on memory of two important factors in the lifetime experiences of the child: formal schooling and urban environment. The selection of memory tasks was of considerable importance. The two tasks, short-term and recognition memory, were selected because: (a) Each task tapped both control processes and structure in memory, and therefore, each could complement (or contradict) the other; (b) a large Western literature exists for both tasks: and (c) both tasks are pictorial, require little verbal response, and are therefore easy to administer to individuals of widely varying background.

The remainder of this paper is organized into five sections. The first of these provides a detailed discussion of the Moroccans who participated in the experiments. The reader interested in a more thorough background concerning these subjects and childhood in Morocco should refer to Wagner (Note 4,1977). Experiment I, a study of short-term memory follows, where emphasis is placed on the development of control

MEMORIES OF MOROCCO 5

processes. A study of recognition memory appears in the following section, Experiment II, where emphasis is placed on structural features in memory. The important results of the two studies are summarized in the fourth section; and, in the final section, are discussed in terms of the underlying motif of this paper: Control processes seem to develop as a function of certain childhood or lifetime experiences, while structural features appear to be more invariant over age.

MOROCCAN SUBJECT POPULATIONS

The present study was undertaken in Morocco primarily because Morocco provides a large range of populations and environments which could be useful in comparing and contrasting the effects of schooling and environment on development.

The subjects in this study were all Moroccan males; Moroccan females over 12 years of age could not participate due to Islamic custom. Subjects were selected so as to complete a three-way factorial design: age (4) x school (2) x environment (2). Over an age range from about 6 to 22 years, this design made it possible to study the independent effects of both schooling and environment on memory development. The age and schooling characteristics of the main groups used in the study are presented in Table 1; and each factor will be discussed in some detail.

The ages of urban schooled and rural schooled subjects were determined by documents provided by school authorities, which were originally based on birth documents. In all cases where adequate documentation (such as birth documents) was not available (mostly among rural nonschooled subjects) other methods were used to ascertain age. The most useful was to gather information from relations and friends of the subject pertaining to relative birth years and birth orders.

Schooling

Modern public schools in Morocco, both urban and rural, were originally developed along the French education model. Morocco currently has a national school policy in which curriculum materials and training of teachers are controlled in all parts of the country, producing a system where remarkably few differences exist between the urban and rural school settings selected for this study.

Urban schooled subjects were selected from two grades of primary school and two grades of secondary school in the city of Marrakech. All possible subjects from a single classroom were selected so as to avoid teacher-biased selection. Comparable rural schooled subjects were

6 DANIEL A. WAGNER

TABLE I

MOROCCAN SUBJECTGROLJPSBY YEARSOF AGE ANDSCHOOLING

Age Schooling

Group

Urban/Schooled (US)

N Mean Range Mean Range

24 6.97 6-7 1.00 I 24 10.50 10-12 3.08 3-4 24 13.37 13- 14 6.96 6-8

24 18.42 18- 19 11.46 IO- 14

Rural/Schooled

(RS)

24 8.12 8-9 24 10.12 10-11 24 13.65 13-14 24 18.96 18-21

Urban/Nonschooled

(UN)

24 7.54 6-8 24 10.79 10-12 24 14.50 13-16 24 18.92 17-22

1.00 I 3.17 3-4 7.54 6-10

11.58 IO-14

.04 o-1

.08 o-1

.04 o-1

.08 o-1

RuraVNonschooled 24 7.12 6-8 .08 O-l (RN) 24 10.71 10-12 .17 O-l

24 14.33 13-16 .17 O-l 24 19.79 17-22 .08 O-1

Koranic students 24 19.42 18-25 2.00 l-6 Rug sellers 12 26.42 17-36 3.33 o-7

obtained from two grades of primary school and two grades of secondary school in two rural villages in the Middle Atlas mountains. The primary school was in a village considerably more traditional than the village with the secondary school (see comments on environment below). However, since many students in the secondary school came from the more traditional surrounding villages, it was possible to select a majority of these students from those who earlier had gone to the same or similar primary schools in the region.

Environment

The urban environment was the city of Marrakech (pop. 250,000) in the south central plain of Morocco. Marrakech, as with most urban centers, has many cinemas, much written and pictorial materials, and public television is broadcast in many cafes where men gather at various times in the day and evening. Although many Marrakech residents are of either Berber or Arabic linguistic heritage, most of the urban subjects tested spoke Moroccan Arabic as their maternal (or first) language.

MEMORIES OF MOROCCO 7

The rural setting was in the Middle Atlas mountains region of the Province of Beni-Mellal. The primary school for testing younger rural schooled children was in the village ofArhbala (est. pop. lOOO), about two hours’ drive by difficult mountain roads from the closest village, El Ksiba (est. pop. 3000), which has permanent telephone and electricity links to modern urban Morocco. This latter village had a secondary school where older rural schooled subjects were tested.

Rural nonschooled groups of subjects were obtained on a paid volunteer basis from Arhbala’s smaller surrounding villages, with populations ranging from 200 to 500 inhabitants. The rural economy is primarily based on small agricultural and pastoral efforts. A few of these subjects had attended school briefly (for about 1 year), usually while visiting a relative in Arhbala or another similar village. It appeared that rural nonschooled subjects did not go to school for the simple reason that schools were unavailable within reasonable traveling distance from their homes.

Families of rural schooled subjects were, in general, better off economically than families without children attending school, although this was not universally the case. The maternal language of almost all rural subjects was Berber. Rural schooled subjects could speak Arabic after a few years of school training, although many took longer to learn to read and write. Most rural nonschooled subjects could neither speak nor read and write Arabic.

Two additional groups of Moroccans were also studied. Korunic

students, the first group, were tested because they were thought to have special memory skills as a function of their preoccupation with memorization (cf. Hardy & Brunot, 1925). Koranic students were obtained at a traditional school of Koranic scholarship in the town of Sidi Zouine (est. pop. lOOO), about 30 miles from Marrakech. This relatively poor rural village was built around the school founded by a famous Koranic scholar descended from the family line of the Prophet Mohammed. Koranic students spent most of their day and night studying and memorizing parts of the Koran. They typically spent all day in the school--the mornings with a teacher, and the afternoons studying and memorizing what the teacher said or wrote on the wooden slabs that served as portable “blackboards.” Methods of memorizing appeared to involve writing special notes or commentaries next to certain text passages, and vocal rehearsal of passages. This latter method was sometimes accompanied by changing tonalities in chanting that appeared to mark segments of the passage being memorized.

The second group was composed of 12 Rug Sellers from the Marrakech rug market. As with the Koranic students, the Rug Sellers were selected and tested because it was thought that they might have developed special memory skills particularly well suited for remembering rug designs (which were the stimuli in the recognition memory task in Experiment II). The

typical rug seller in the rug market participated in a daily wholesale auction

8 DANIEL A. WAGNER

of rugs brought in from various rural areas around Marrakech. Rugs were carried around to each seller, who made repeated bids on certain rugs throughout the day. During this period, the rug seller had to remember not only his bid, but also the various rugs for which he had made each bid. Thus, the rug seller had to remember or recognize up to perhaps a dozen or more different rugs each day, and which changed from day to day.

EXPERIMENT I: SHORT-TERM MEMORY

A considerable body of research has shown that age-related increases in performance on short-term memory tasks involve the progressive utilization of mnemonic skills or strategies by the individual. In serial memory tasks, most of the age-related changes occur at the primacy or early portions of the list to be remembered, and this is said to be a function of memory strategies (Belmont & Butterfield, 1969; Brown, 1975; Cole, Frankel, & Sharp, 1971; Hagen, Meacham, & Mesibov, 1970).

Most of this research has been based on American or Western children, for whom age, education and urban environment are almost perfectly correlated. It has generally been impossible, therefore, to determine which factors or combinations of factors might be responsible for developmental changes in memory. As mentioned earlier, Wagner (1974), working in Yucatan, found that chronological age alone could not account for typical patterns in the development of short-term memory. One serious problem in the Yucatan study was the confounding of schooling and urban environ- ment. Older schooled children lived in urban Merida (the provincial capital of Yucatan), while nonschooled children lived in the rural village of Mayapan. Either schooling or urban environment could have produced increased memory performance.

Experiment I was designed to separate the effects of schooling and environment, and to replicate, in another culture, the previous findings concerning the effects of schooling on short-term memory development.

Method

Subjects. The subjects tested in this experiment were: urban schooled, rural schooled, urban nonschooled, rural nonschooled, and Koranic school students.

Stimuli. Test materials were adapted from Wagner (1974). The stimuli were sets of seven white cards (3 x 5 in) with two line drawings drawn on one side (see Fig. 1). Each ofthe seven cards had a particular object and a particular animal on it (one above the other: three with animals above, and four below). The line drawings were separated from each other by a 1 cm space. Each animal was paired with one object as follows: fish-hammer, dog-earthen jug, sheep-cup, camel-knife, goat-book, chicken-key, cow-shoe. Two items were presented on each card so that a subsequent test for incidental memory for pairs could be conducted; the incidental memory data are discussed in Wagner (Note 4). Pretests showed that the line drawings, which were chosen to be familiar to Moroccan children, were easily recognizable by all subjects.

Fourteen sets of seven stimulus cards were constructed and arranged in a fixed randomized

MEMORIES OF MOROCCO 9

FIG. 1. Set of stimuli used in the serial short-term memory task. The rooster is the “probe” stimulus that is presented following the presentation of the seven cards-to-be-remembered.

order, such that each serial position could be tested twice. A similar, but smaller set of stimuli, consisting of three animals and three objects, was constructed for use in the practice session that preceded the standard task. A set of 14 stimuli contained single drawings of animals, and these were used as probe stimuli to test serial short-term memory.

Procedure. All subjects were tested individually by a single trilingual Moroccan experimenter, who conducted the testing in the maternal (first) language ofeach subject, with the exception of older schooled subjects, who were tested in Moroccan Arabic. The experimenter first interviewed each subject and completed a questionnaire which provided demographic information, served as a warm-up period for the subject, and allowed for control of selection. The place oftesting was usually an unused classroom for the school students, and ranged from an unused room in a community center to a Berber nomad’s tent for rural nonschooled subjects.

The task required remembering the locations of the seven animal drawings as each was presented briefly (2 set) and placed face down in front of the subject. The cards were presented one at a time from right to left3 forming a row of seven cards. Then, after about 2 seconds, the subject was shown a probe card with a single predetermined animal drawing on it, and was required to point to the card in the row that contained the same animal. The experimenter, who sat opposite to the subject at a table or a desk, first gave six practice trials

3 In contrast to the Yucatan study (Wagner, 1974), where the stimuli were placed from left-to-right, in the Morocco study, stimuli were presented from right-to-left, in accord with the direction by which literate Moroccans read Arabic.

Only animal stimuli were used as the “relevant” items to be remembered, because it was

found in an earlier study (Wagner, 1974) that neither group of stimuli had any special effects on performance.

10 DANIEL A. WAGNER

of three cards per row. If necessary, he mentioned the names of the animals and objects. The subject was instructed to try to remember only the animals, and that the objects were “not important” (for an equivalent and more detailed set of task instructions, see Wagner, 1974). The 14 test trials were arranged randomly so that no picture appeared next to the same picture in the array of rows presented.

Performance for the serial short-term memory task was assessed for two scores: proportion correct out of 14 trials, and proportion of correct responses for each serial position. Also, a middle-positions score (mean of positions 3, 4, and 5) was calculated to provide a measure with less ofthe presumed effects of specific memory skills which enhance performance at both the primacy and recency portions of the serial position curve (Wagner, 1974).

Younger subjects were rewarded with candy or small amounts of money at the beginning of the experiment (which also consisted of several other cognitive tasks). Older schooled subjects were required by school authorities to participate, and therefore, received no special rewards. During the practice trials, less than 1% of the subjects failed to achieve better than chance performance, and these were eliminated from further testing, and were replaced by other subjects.

Results

Three-way analyses of variance, age (4) x school (2) x environment (2), were performed on total recall (shown in Fig. 2), and on primacy (serial position I), recency (serial position 7) and middle-positions scores, as shown in Fig. 3. Not present in the figures are the data from the extra group of Koranic students, who obtained the following scores: total correct = .42; primacy = .49; recency = .63; and middle-positions = .35. Results of the analyses of variance appear in Table 2. The most important features of these analyses may be summarized as follows:

.70-

E

c

c .50 -

z

E

f? 0 E

.30 -

- us - us - RS - RS

* ._........ * * ._........ * UN UN

q ._........ q q ._........ q RN RN

. . . . ..a . . . . ..a

/.... /’ /.... /’ ..O- . . . . . . . ..__......_ q ..O- . . . . . . . ..__......_ q _...... N”. _...... N”.

q . ...‘” q . ...‘”

b I I I 7 10 14 19

AGE

FIG. 2. Total short-term recall (summed over serial positions). US = Urban/Schooled; RS = Rural/Schooled: UN = UrbaniNonschooled; RN = RuraliNonschooled.

MEMORIES OF MOROCCO 11

l.OOr

2 1.00

8 <*A ,__._... 3 $::..L I ,.__..___ ... . . ..__. t ._..

1 -

.,... . ..__. rl __._______._.__.... (1

I Recency I I I 1

-us

- RS

l l UN

u......~ RN

Middle-Positions 0

7 10 14 19 AGE

FIG. 3. Primacy, recency, and middle-positions recall.

1. All main effects were significant for total, primacy and middle- positions recall. For recency scores there was a small but reliable effect for school. It should be noted that the age effects for primacy and total recall were considerably larger than that of middle-positions.

2. The age x school interaction was significant only for primacy and total recall.

3. Three school x environment interactions were significant. Primacy and total recall performance were enhanced in the urban schooled environment, while recency recall was greater in the rural schooled environment. Furthermore, enhanced recency occurred mainly for the younger rural schooled children (as indicated by the significant age X location interaction).

Simple comparisons for total recall were made in order to further specify differences between relevant groups. These analyses indicated that, at younger ages (7 to 10 years), all groups performed similarly, except for the low scoring rural nonschooled groups [t(46) s 2.28,~ < .05]. At older ages (14 to 19 years), both rural and urban schooled groups, which did not differ

12 DANIEL A. WAGNER

TABLE 2

ANALYSIS OF VARIANCE (F SCORES) FOR SHORT-TERM RECALL MEASURES

Source df Total Primacy Middle-positions Recency

Age (A) School (S) Environment (E) AxS AXE SxE AxSxE

* p < .05. **p < .Ol.

3, 368 22.06** 15.88** 7.22** 2.21 1, 368 60.87** 14.21** 22.06** 5.88* 1, 368 35.28** 14.21** 19.78** .02 3, 368 8.75** 10.62** 2.01 .33 3, 368 .91 .32 1.21 3.29* 1, 368 13.88** 10.98** 1.97 9.72** 3, 368 1.53 3.75* 3.32* .65

from one another, performed better than the nonschooled groups [t(46) 2 2.08, p < .OS]. At age 19, each urban group scored better than its rural counterpart [t(46) 3 1.70, p < . lo]. Finally, Koranic students performed most like nonschooled subjects [t(46) c 1.70, p > . lo], while differing significantly from both schooled groups [t(46) 2 3.35, p < .Ol].

Simple comparisons within primacy recall showed that: There was no improvement in performance in any group from ages 7 to 10 [ t(46) c 1.58, p > . lo]; performance improved among both schooled groups from ages 10 to 14 years, [t(46) 2 3.12, p < .Ol]; and the younger urban nonschooled subjects (7 and 10 years) performed slightly better than the other same-age groups, [t(46) > 1.81,~ c .07].

Further comparisons revealed that there was essentially no primacy effect (primacy over middle-positions) for all the younger age groups (7 and 10 years) [t( 190) = .72, p > . lo], nor for the older nonschooled groups (14 and 19 years) [t(94) = .75, p > .lO]. However, the primacy effect was significant for the older schooled groups [t(94) = 5.63, p < .Ol].

Discussion

While schooling and urban environment were confounded in the Yucatan study, from the present study it appears that both schooling and urban environment enhance the development of short-term recall. Analyses of total recall produced a number of interesting effects. There were no differences among the younger children who were either in school, in an urban environment, or both, whereas rural nonschooled children performed considerably less well. Until the present study, it had remained unclear whether the lack of differences between younger Mayan and Mestizo children in Yucatan was due to similarity of the schooling expe- rience for both groups or to a more general maturational factor. In Morocco, younger urban children performed similarly, while rural non- schooled children performed less well, showing that the urban environ-

MEMORIESOFMOROCCO 13

ment was sufficient for good performance at these ages. Rural schooled subjects performed much like the two urban groups, showing that school- ing was sufficient for good performance in the absence of an urban environment.

However, the results for the older age groups indicated other interesting relationships. Children in school performed better than nonschooled children, but only at older ages (over 13 years). Therefore, while the urban environment might be sufficient for effective development of short-term memory at younger ages, it seems insufficient for older subjects. However, among nonschooled subjects, the older urban subjects still maintained their earlier performance advantage over rural subjects. More- over, the urban environment provided some additional advantage over the rural environment for older subjects who were in school.

In general, then, short-term recall of Moroccan children, from the earliest age groups, appeared to benefit from something in the urban environment, while the effects of schooling increased at older ages.

Can we identify the within-trial sources of the varying effects of age, schooling, and environment? Recency recall appeared least affected by these variables, as only a moderate main effect for schooling was found. The recency effect did not drop below a mean of 60% for any group, which was higher than any group’s middle-positions recall. Middle-positions recall appeared qualitatively similar for all groups, although each main effect was significant. Primacy recall produced interactions that appeared most characteristic of total serial recall. As in total recall, primacy increased with age only when coupled with schooling and this effect was enhanced by the urban environment.

The increase in primacy in the schooled groups is consistent with the hypothesis that verbally mediated rehearsal strategies are more available to older, more experienced children. The implication is that rural non- schooled and to some extent, urban nonschooled subjects, showed little developmental improvement in primacy and overall recall because they were not engaging in these mnemonic strategies. While the relationship between strategies and primacy recall has been studied in a variety of earlier studies, it is necessary to make an important qualification here. Systematic measurement of verbal rehearsal was not undertaken in the present experiment, although informal observations indicated that some schooled children appeared to rehearse aloud, while this was seldom the case among nonschooled children. Two pieces of evidence have consistently supported the relationship between verbal rehearsal and primacy effects. First, direct observation of children’s rehearsal strategies has shown a significant positive relation between rehearsal, primacy effects, and increased recall (Flavell, 1970; Hagen, 1971). Second, children who have been induced or trained to use rehearsal skills have shown increased short-term recall (Keeney, Cannizzo, & Flavell, 1967), and

14 DANIEL A. WAGNER

particularly at the primacy positions (Kingsley & Hagen, 1969). This type of evidence suggests that rehearsal skills, in such a task, are likely to be an important factor in producing the primacy effect and increased recall. However, while such an hypothesis is plausible, other types of strategies might be involved. In any case, it seems clear that some type of school-related strategy-probably rehearsal-was the basis of develop- mental increases in recall for older school children.

As suggested earlier, memory development can be meaningfully discussed in terms of models that distinguish between structure and control processes. With respect to the results just presented, two points may be suggested: First, recency recall or short-term store is present in stable form in all population groups studied, although the size of the effect may vary by group; and second, control processes, such as verbal rehearsal, increase developmentally only for schooled subjects, and in somewhat diminished form among urban nonschooled subjects. While earlier studies have typically confounded schooling and urban environment, the present study showed that each experiential or cultural factor may have an effect on the development of control processes in memory. Although one might hypothesize from this experiment that control processes increase only as a function of years of schooling, such an assertion is untenable because chronological age-related changes may occur simultaneously and may interact with task performance.

It has been assumed by certain investigators (e.g., Cole, Gay, Glick, & Sharp, 1971; Jenkins, 1974) that cognitive skills develop best in the context of regular use. Under this assumption, Koranic students, who spend most of their day memorizing and apparently rehearsing aloud the Koran, might be expected to perform as well as, if not better than, the other subjects in this study. In fact, these Koranic students performed much like nonschooled subjects, showing little evidence of rehearsal strategies. An alternative suggestion might be that this task was inappropriate to the kind of stimuli or “content” that Koranic students typically memorize or process.

Although no subsequent recall experiments have yet been undertaken with these Koranic students, observations and tape recordings were made of their manner of memorizing and rehearsing the Koran. An interesting feature of these tapes is that rehearsal takes place in a particular fashion with sections of text marked by high-pitched and louder intonation and sometimes accompanied by a stick being tapped against the Koranic script board (where the text has been written). Analysis of this apparent use of rhythmic marking has not yet been undertaken, although it would provide interesting evidence for development of mnemonics as a function of specific cultural experiences.

EXPERIMENT II: RECOGNITION MEMORY

Developmental studies of recognition memory, in contrast to the development of recall memory, have been most often characterized by a

MEMORIESOFMOROCCO 15

lack of age-related trends in performance. In a recent review, Brown (1975) has suggested that recognition memory for pictures does not require active retrieval or acquisition strategies, and should, therefore, be relatively uninfluenced by the age of the child.

Models of recognition memory (e.g., Anderson & Bower, 1972; Kin&h, 1970) suggest that there are two primary parameters that determine recognition memory performance: acquisition, the amount of information that enters the memory system and may be studied immediately after presentation; and forgetting rate, the continuous decay of information from memory as a function of time or intervening information to be remembered. The forgetting rate, as discussed earlier in the model of Atkinson and Shiffrin (1968), is considered to be a structural feature of memory; acquisition is more variable, and is considered to be a function of many factors such as the type of stimulus encoding and perceptual set.

Invariant forgetting rates seem to be responsible for the lack of age-related changes in recognition memory performance. Studies have shown that forgetting rates exhibit little age related change either for words (Belmont & Butterfield, 1969; Berth & Evans, 1973; Craik, 1971; Wickelgren, 1975) or for pictures (Fajnsztejn-Pollack, 1973; Nelson, 1971; Olson, 1973). Where age-related differences in performance have been found, the acquisition parameter seems to play a major role. In several of the studies mentioned above, acquisition, or immediate recognition memory, appears to increase with age to adulthood for words (Berth & Evans, 1973; Wickelgren, 1975) and for pictures (Fajnsztejn-Pollack, 1973; Nelson, 1971).

The origin of developmental changes in acquisition-and in some cases forgetting rates-may be a function of stimulus variables and the subject’s encoding abilities for different types of stimuli. Of particular importance seems to be the labelability and familiarity of stimuli to the subject. Recent developmental studies, primarily using recall paradigms, have found that semantic encoding abilities develop with age, and are an important factor in developmental differences in memory performance (Hagen et al., 1975). The “depth of processing” model of Craik and Lockhart (1972) suggests that information that is better encoded (i.e., more “deeply processed”) will be retained better in memory. Support for the depth of processing model for picture recognition memory with adults has been provided by Bower and Karlin (1974).

In sum, American or Western studies of the development of recognition memory indicate that forgetting rates, if independent of stimulus encoding variables, are invariant with age, and may be considered to be structural features of memory. In contrast, acquisition parameters seem subject to considerable age-related change. At present, it is unclear the extent to which acquisition is a function of materials to be learned per se, or a function of the developing individual’s ability to encode certain materials better than other materials. or both.

16 DANIEL A. WAGNER

In Experiment II, the same contrasting populations were tested in order to ask two main questions: (1) To what degree is rate of forgetting variable with age or experience?; and (2) To what degree can we specify the nature of variation in rate of acquisition ? In order to delineate the lifetime experiences that lead to variation in stimulus encoding, which seems to affect rate of acquisition, stimuli composed of rug designs were selected to be moderately familiar to Moroccan children, very familiar to a particular nonschooled population (Moroccan rug sellers), and relatively unfamiliar to a very schooled population (University of Michigan undergraduates).

Method

Subjects. Subject groups were composed ofthe same individuals as in the short-term recall task, with the addition oftwo extragroups: Rug Sellers and Michigan psychology students. As discussed earlier, the Rug Sellers were required to remember rugs as a daily occupation, and were included in the present task to determine if such experience would affect performance in recognition memory for rug patterns.

The other additional group was that of 24 undergraduate psychology students at the University of Michigan. Since situational variables (e.g., motivation, experimenter, testing conditions, etc.) varied greatly between Morocco and Michigan, overall performance levels (or acquisition) in the comparison of groups were of less interest than forgetting rates.

Stimuli. The present experiment used a modified version of the continuous recognition memory task of Shepard and Teghtoonian (1961). The stimuli were 207 black and white photographs (5 x 7 in) of Oriental rugs (see Fig. 4). The experiment consisted ofa practice test (30 trials), followed by the experimental task (177 trials). The practice task consisted of 15 original and 15 duplicate rug patterns, which were arranged so that repetitions (i.e., duplicates) occurred at varying intervals or lags in the series. Thus, repetitions of patterns

FIG. 4. Examples of two rug pattern stimuli used in the recognition memory task.

MEMORIES OF MOROCCO 17

occurred at lags ranging from 0 (i.e., duplicate was repeated on the next trial) to 17 (duplicate repeated after 17 intervening patterns).

The experimental task consisted of 88 different patterns and 88 duplicates, with a single extra or filler pattern presented only once. These 177 items were arranged in a sequential array so that duplicates formed lags of 1,5,10, and 25 intervening items. There were 22 repetitions at each of these four lags, which were distributed in a relatively even distribution over the entire sequence of items (lag 25 was necessarily presented with less frequency in the early portion of the series). Both practice and experimental stimuli were arranged in two large loose-leaf notebooks, so that when the next pattern was exposed, it covered the previous pattern.

Procedure. Each subject was tested individually. The procedure for the practice and experimental tasks was identical, with the exception that each subject was given feedback and prompting on the practice trials. The subject was given the following instructions for each task:

These are pictures of rugs. Each rug pattern that you see will have one and only one duplicate (“sister”) copy. Look at each rug very carefully. Tell me whether this is the first or second time you have seen this rug design in the test booklet. I f it is the first time say “first” (or “not seen before”): if it is the second time, say “second” (or “seen before”).

The subject was allowed about 5 seconds to look and respond to each item before turning to the next item. The experimenter occasionally had to go through the practice task in detail with some of the younger, particularly nonschooled, children. About 3% of all subjects were eliminated from the sample for lack of task comprehension. All Moroccan subjects were tested by a single trilingual Moroccan experimenter, in the manner described earlier. American subjects were tested in their native language by a research assistant at the University of Michigan. Testing time was approximately 45 minutes for each subject. Moroccan subjects received small amounts of candy or money at the beginning and end of the test period to ensure motivation.

Results

Five measures of recognition memory were derived from the continuous recognition memory task: total correct4 (the sum of hits and correct rejections for each subject; see Fig. 5); and d’ for each lag (1, 5, 10, or 25 intervening items; see Fig. 6). For each of these measures, three-way analyses of variance, by age (4) x school (2) x environment (2), were performed and are presented in Table 3. The most important features of these analyses may be summarized as follows:

1. Chronological age produced little or no reliable effects for the various recognition measures.

4 Total correct is the sum of hits [P(old/old)] and correct rejections [P(new/new)]. In the present discussion, old refers to “second,” and new refers to “first,” as in the task instructions. Mean d’ (averaged over the four lags) was not used as an alternative measure to total correct, because higher mean d’ values at shorter lags would tend to bias the resulting score towards these lags.

The d’ score is generally calculated as the z-score distance separating two theoretically normal distributions of new and old items (for more details, see Banks, 1970). A recent and simpler method of calculating d’ was derived by Smith (Note 3). This formula produces ad’ value that never varies more than about 1% from the z-score values of earlier formulas, and was used for all (I’ calculations in the present study.

18 DANIEL A. WAGNER

85- *

u’ >

ii q

i2 759 / 7 2---p

_..’ . . . [ , _..___........ [ ] _..’

z

i ii -“* *M

65 - --* wR *......* “N : ______,! RN CIK

I I I I 7 10 14 19

AGE

FIG. 5. Total correct recognition (sum of hits + correct rejections). Extra groups: M=Michigan; R=Rug Sellers; K=Koranic students.

2. Schooling produced significantly increased performance for the longer lags (10 and 25 intervening items), which resulted in a significant schooling effect for total correct.

3. The effect of environment was highly significant, Contrary to the findings in the short-term recall task, the rural subjects, whether schooled

3.Or

l,LAG,lO, ,

7 10 14 19 7 10 14 19

AGE

FIG. 6. Mean d’ values for lags 1, 5, 10, and 25 intervening items.

MEMORIES OF MOROCCO 19

TABLE 3

ANALYSIS OF VARIANCE (F SCORES) FOR RECOGNITION MEMORY MEASURES

Source df Total d’(l) d’(5) d’(10) d’(25)

Age (A) School (S)

Environment (E) AxS AXE SxE AxSxE

3, 368 2.16 3.66* 3.02* 2.80* 1.41 1, 368 9.28** .Ol .83 7.68** 20.43**

1, 368 71.06** 66.23** 57.8.5** 45.09** 18.56** 3, 368 2.51 2.45 4.97** 1.58 5.92** 3, 368 .I6 2.16 .98 .39 .71 1, 368 .36 7.79** .67 2.22 3.96* 3, 368 .15 .51 .38 .18 .25

* p < .05. ** p < .Ol.

or nonschooled, performed significantly better than their urban counter- parts on all recognition measures.

Forgetting rates-or the decrease in d’ over lag or delay-were also of interest. The forgetting curves are plotted by group in Fig. 7. Profile analyses5 were performed within each of the groups, and the results confirmed the observation that the forgetting curves were invariant (i.e., parallel) with age for all groups (maximum root = . 15;~ > .05), except the urban schooled subjects (maximum root = .41; p < .Ol), where some age differences in forgetting rate occurred. However, even in the urban schooled groups, the change in forgetting favored neither a monotonic increase nor decrease in rate with age.

Since there were generally no age differences in forgetting rates, these data were pooled across ages and were compared, as single groups, with the data from the three extra groups (Rug Sellers, Koranic, and Michigan subjects) in Fig. 8. Profile analyses of these curves indicated that there were no differences in forgetting among urban nonschooled, both rural schooled and nonschooled groups, Koranic, and Rug Seller groups (maximum root = .l 1; p > .05). Urban schooled and Michigan subjects showed less forgetting over lags, and were both significantly different from the other five groups (p 2 73.65; ci’ = 3,440;~ < .Ol), and were different from one another (TL = 17.43; cif = 3,116; p < .Ol).

d Profile analyses (Morrison, 1967) were used to compare the forgetting rates because they require no assumption of linearity. Significant “rnaximurn root” or T’ values imply that the curves are nor parallel. The parameters for the three maximurn root calculations cited in the text were, respectively: 3, -.5,44; 3, -.5,44; and 3,0.0,49.5. Logarithmic transformations have been used by some investigators (e.g., Wickelgren, 1975) to produce linear functions of d’, testable by regression analyses. In the present case, however, there were about 20 negative d’ values (i.e., at about chance level performance) for which log d’ could not be calculated. The profile analyses were carried out within the MIDAS computer program (University of Michigan), and the author would like to express his appreciation to Ken Guire for his helpful suggestions with the statistical analyses.

20 DANIEL A. WAGNER

2.0 -

-- 1.5 -

1.0 -

RN 'b... “.,:’

so ., -

2.5-

2.0-

'0 1.5 -

1.0 -

.50.’ 1 5 10 25

LAG

FIG. 7. Forgetting curves (mean d’ over Iags) for each main group overages 7, 10,14, and 19 years.

2.5 -

~~---.~ RN m--EI R

es--a K

1 5 10 25 LAG

FIG. 8. Forgetting curves (mean d’ over lags) for each subject population (summed over age).

MEMORIESOFMOROCCO 21

Additional simple comparisons were performed to evaluate the three extra groups (Rug Seller, Koranic, and Michigan subjects). In terms of total correct recognition, when compared to other groups at age 19, Michigan subjects performed better than all other groups [r(34,46) 2 2.81, p < .Ol]; Rug Sellers performed better than all groups except Michigan and rural schooled subjects [t(34) 2 1.92,~ 6 .05]; and Koranic students performed at the level of rural nonschooled and both urban schooled and nonschooled subjects (p > .lO).

Discussion

In general, the present data support previous research that indicated little age-related change in forgetting rates for pictorial recognition memory (Brown, 1975). With the exception of two age groups within the urban schooled sample, Moroccan subjects showed similar or parallel forgetting rates across age. Furthermore, forgetting rates were also invariant with respect to schooling and environment. As shown in Fig. 8, five out of seven different groups showed parallel forgetting rates.

Urban schooled Moroccans and Michigan subjects had flatter or lower forgetting rates than any of the other groups tested. As structural features of memory, forgetting rates were hypothesized to be invariant over all subjects, assuming similar stimulus encoding of the materials. Thus, one might hypothesize that these two anomalous groups may have encoded the rug stimuli differently than did the other groups. Some evidence for such an hypothesis was gathered in posttest interviews with the Michigan subjects, who claimed to use various mnemonics to aid in subsequent recognition, such as: counting the number of discrete shapes in the rug patterns; comparing the differences in rectangularity of the patterns; and attempting to gauge the relative brightness of the photographic prints. It seems possible that the modern, media-oriented groups-i.e., urban schooled Moroccan and Michigan subjects-might have used more elaborated stimulus encodings (as in Bower & Karlin, 1974, mentioned earlier).

Overall, or total correct, recognition performance was best for rural and schooled subjects, while age, again, was not an important factor. In fact, rural nonschooled subjects scored higher in total correct than urban schooled subjects, who had performed best in the short-term recall task discussed in Experiment I. It is unclear why rural subjects performed so much better than urban subjects on the recognition memory task. One potential hypothesis for this difference might be due to differential familiarity with rugs between the two environments, but this seems doubtful, given the widespread contact with, and usage of, rugs in both urban and rural regions of Morocco. Furthermore, in one of the few tests of familiarity effects on children’s recognition memory, it was found that Guatemalan children recognized familiar and unfamiliar pictures with equal facility (Kagan, Klein, Haith, & Morrison, 1973).

22 DANIEL A. WAGNER

With respect to the three extra groups, Michigan and Rug Seller subjects showed enhanced recognition performance, while Koranic students did not. As noted earlier, Rug Sellers worked continually with rugs, and had to remember rug patterns for their trade. It seems plausible that these Rug Sellers, as a function of occupational experience, were so familiar with the general pattern or form of rugs that they were able to pick out the distinctive features in the patterns with greater famility than other subjects. Such an hypothesis would fit with the perceptual learning of distinctive features in the theory of Gibson (1969).

How does one explain the superiority of Michigan subjects in overall performance on this task? One explanation could be that Michigan subjects performed in ideal situational testing circumstances, such as a quiet, well-lit room, and on a task that was within the typical cultural behavior of psychology undergraduates. An alternative explanation might be that Michigan subjects performed best because they were best at perceptual analysis, as a function of schooling, mass-media, child-rearing patterns (in the “cognitive style” or “field independence” theory of Witkin, 1967). A third possibility might be that these college sophomores were a very select group of subjects (they were probably in the top 20% of the American population on U. S. intelligence tests), and, therefore, a general intelligence factor might be an important variable in their high perform- ance. Finally, it should be noted that the Michigan data were included here only as a familiar “landmark” to which the Moroccan data might be compared-but strict comparison across cultures is hazardous given the factors enumerated above.

In conclusion, it should be emphasized that the rural children did extremely well on a memory task of considerable complexity. The rug designs were much more complex than the simple animals and objects that are often used in children’s memory tasks. We have also seen that the rural nonschooled subjects performed better than the urban school subjects. This result is especially important because it indicates that rural nonschooled subjects-the most “traditional” of the groups tested- could understand, attend to, and perform well on such cognitive tasks. It seems likely, therefore, that the performance of these same individuals on the earlier short-term recall task was not a function of mere disinterest or lack of motivation in the testing situation. The different effects of the urban and rural environments on the two memory tasks-for which no simple interpretation is offered here-illustrates the complex nature of the relationships between particular tasks, particular cognitive skills, and particular aspects of culture.

SUMMARY OF THE MAJOR FINDINGS OF EXPERIMENTS I AND II

Before attempting to discuss the results of the two memory studies, it is useful to review the major findings. These may be summarized as follows:

MEMORIES OF MOROCCO 23

1. Short-term store (recency effect) was present and relatively invariant by age, schooling, and environment.

2. Verbal rehearsal (primacy effect) was present in only the older (from about age 13) schooled subjects. There was also a small, but significant effect for children in the urban environment. Koranic students, originally thought to possess special memorizing or rehearsal skills, performed at a relatively low level, much like rural nonschooled subjects.

3. Overall recall on the short-term memory task was greatest for older schooled subjects, and this seemed to be a direct function of increasing primacy recall with increasing age and schooling. Urban nonschooled subjects showed little improvement with increasing age, and performed better than their rural nonschooled counterparts. This appeared to be a function of enhanced primacy recall.

4. In the recognition task, forgetting rates generally were found to be constant over age, and similar over five of the seven groups tested. The two groups (urban schooled Moroccan and Michigan subjects) who showed lowest forgetting rates may have been using some type of “deep processing” or encoding strategies.

5. Recognition memory performance (total correct) was greater as a function of rural environment and schooling. However, Michigan, Rug Seller, and rural subjects showed the best overall performance (in that order).

CONCLUSIONS

Models of memory postulate a set of features that includes both structure and control processes. It was suggested earlier that structure in memory may mature very rapidly in young children. It was further argued that such structure should, if built-in, be evident in all individuals regardless of age or experiential background, such as schooling and environment. Several pieces of evidence support this hypothesis: (a) The recency effect was found in all groups, regardless of age or background; and (b) forgetting rates were generally invariant across groups.

Developmental research in memory has shown that control processes-such as verbal rehearsal and clustering-improve between the ages of 5 and 15 years. While chronological age or maturation has been said to be the important independent variable in such research, some earlier cross-cultural studies (Wagner, 1974, 1975; Cole, Gay, Glick & Sharp, 1971) have shown that the development of control processes may be dependent in part on formal schooling. Data from the present study adds further support to the hypothesis that experiential factors, such as schooling and living in an urban environment, influence the development of control processes. The results of the short-term recall task showed that verbal rehearsal appeared to be used only by older schooled subjects, and to some extent by urban nonschooled subjects. These data, reflecting the

24 DANIEL A. WAGNER

use of verbal rehearsal strategies by about age 13, are consistent with data collected among American school children (Hagen, 1971). As mentioned earlier, in this and many previous studies, the rehearsal strategy was not directly measured. However, it is probably fair to conclude that the strategies that produce memory development are a function of the schooling process rather than merely chronological age.

The complexity of the topic of control processes may be illustrated by the relatively poor performance of the Koranic students in the recall task. In this task, high performance seems to require the use of verbal rehearsal of discrete labelable items for remembering spatial location. Memory of this sort has been termed “episodic” by Tulving (1972), and may be contrasted with “semantic” memory which makes use of knowledge or schemas (as in Bartlett, 1932; or Piaget, 1968) to reproduce or reconstruct the information to be remembered. These Koranic students were originally thought to have special memory abilities, either by their selection for study at the Koranic school, or as a function of daily memorization of the Koran. The Koranic students, despite frequent statements about such skills as rote- memorization and verbal rehearsal, may be using these skills only in tasks that are more “semantic” than “episodic.” That is, Koranic students may have well-developed schemas for embedding new information about the Koran, while such schemas are relatively nonfunctional for the processing of spatially presented pictures, as in the recall task. In contrast, school children may often memorize a large array of more-or-less meaningful items, and may have developed skills for episodic memory to a greater extent than Koranic students.

There is evidence to suggest that control processes or mnemonics are culture-specific -that is, a function of particular environmental and cultural experiences. Certain cultures or subcultures seem to have developed techniques for remembering that are not found universally, and that may be particularly useful for the type of information to be remembered on a given task. Thus far, we have seen that Western-style schooling seems to favor the processing of episodic recall in various societies, while Koranic schooling does not. Tape-recordings of Koranic students appeared to show use of a form of rhythmic marking to segment longer verses of the Koran. Such marking, although not yet studied in detail, would appear to be a culture-specific mnemonic. Additional evidence for culture-specific mnemonics is available from other sources. It is commonly known that remembering the words to a song is greatly facilitated by singing the song-the tune and rhythm serve as mnemonics. Among the Kpelle in Liberia, Lancy (Note 1) has reported a similar phenomenon, but adds further that “my informants had a great difficulty recalling the songs unless they were singing and dancing” (p. 9). Thus, we see a motoric or kinesthetic mnemonic that aids in recall. Similar evidence was recently gathered in a study of memory in deaf children (Liben & Drury, 1977). In this study, deaf children created their own, apparently

MEMORIES OF MOROCCO 25

culture-specific or deaf-specific, mnemonics for remembering. The authors observed the use of fingerspelling and the use of mime representations (such as “rocking to represent a curved line”) as mnemonics in a short-term recall task, Still other examples of memory aids are numerous in the anthropological literature, and have been summarized in several reviews (Cole & Gay, 1972; Meacham, 1972; and Yates, 1966).

The distinction between built-in versus culturally acquired aspects of behavior is as old as the science of psychology. In many areas of psychology the controversy concerning the relative influences of maturation and environment continues unabated. In a recent and highly publicized study, Kagan and Klein (1973) claimed that the effects of early environmental deprivation had little longterm consequences for children’s cognitive development in Guatemala. In their report, Kagan and Klein concluded:

Our observations in these Indian villages have led us to reorder the hierarchy of complementary influence that biology and environmental forces exert on the development of intellectual functions that are natural to man. Separate maturational factors set the time of emergence of those basic abilities. Experience can slow down or speed up that emergence by several months or several years, but nature will win in the end. The capacity of perceptual analysis, imitation, language, inference, deduction, symbolism, and memory will eventually appear in sturdy form, for each is an inherent competence in the human program. (p. 949; emphases added).

It appears reasonable to conclude that claims such as those made by Kagan and Klein (1973) are limited to the relatively simple estimates of the memory performance they studied. Stated in its strongest form, the present study supports the hypothesis that the structure of memory is a universal in cognition, while control processes seem to be more culture-specific, or a function of the particular experiences that surround each growing child. While the pattern of results appears to support this hypothesis, it is obviously difficult to claim a completely universal structure of memory, because: (1) Only certain structural features of memory were studied; and (2) behavioral universals-like behavioral theories-are impossible to prove. Additional research might find qualitative differences in structural features of memory in other cultures or with other types of tasks. With respect to control processes, large differences appear to be found as a function of children’s lifetime experiences. However, we cannot claim that children growing up in some cultures are unable to use certain control processes or that such processes do not exist in some cultures, for the present study has dealt with the kinds of control processes used on specified tasks. Also, it is important to note that even though an individual may not spontaneously use such control processes, it appears relatively easy to train individuals to employ these processes. In sum, by confirming the development, and the invariance, of several differing aspects of human memory, the present cross-cultural evidence not only seems to support and

26 DANIEL A. WAGNER

complement current models of memory, but also indicates the powerful influence of antecedant cultural and experiential factors.

REFERENCES

Anderson, J. R., & Bower, G. H. Recognition and retrieval processes in free recall. Psychological Review, 1972, 79, 97-123.

Atkinson, R. C., & Shiffrin, R. M. Human memory: A proposed system and its control processes. In K. W. Spence & J. T. Spence (Eds.), The psychology of learning and motivation: Advances in theory and research. New York: Academic Press, 1968. Vol.

BaddeIty, A. D. The psychology of memory. New York: Basic Books, 1976. Baddeley, A. D., & Hitch, G. Recency re-examined. In S. Dornic (Ed.), Attention and

performance VI. Hillsdale, N. J.: L. Erlbaum, 1977. Banks, W. P. Signal detection theory and human memory. Psychological Bulletin, 1970,74,

81-99. Bartlett, F. C. Remembering: A study irz experimental and social psychology. London:

Cambridge University Press, 1932. Belmont, J. M., & Butterfield, E. C. The relation of short-term memory to development and

intelligence. In L. P. Lipsett & H. W. Reese (Eds.),Advances in child development and behavior. New York: Academic Press, 1969. Vol. 4.

Berth, D. B., & Evans, R. C. Decision processes in children’s recognition memory. Journal of Experimental Psychology, 1973, 16, 148-164.

Bower, G. H., & Karlin, M. B. Depth of processing pictures offaces and recognition memory. Journal of Experimental Psychology, 1974, 103, 751-757.

Brown, A. L. The development of memory: Knowing, knowing about knowing, and knowing how to know. In H. W. Reese (Ed.), Advances in child development and behavior. New York: Academic Press, 1975. Vol. 10.

Bruner, J. S., Olver, R. R., & Greenfield, P. M. Studies in cognitive growth. New York: J. Wiley & Sons, 1966.

Chi, M. T. H. Short-term memory limitations in children: Capacity or processing deficits? Memory & Cognition, 1976, 4, 559-572.

Cole, M., Frankel, F., & Sharp, D. Development of free recall learning in children. Developmental Psychology, 1971, 4, 109-123.

Cole, M., & Gay, J. Culture and memory. American Anfhropologist, 1972, 74, 1066-1084. Cole, M., Gay, J., Glick, J., & Sharp, D. The cultural context oflearning and thinking.New

York: Basic Books, 1971. Craik, F. I. M. Age differences in recognition memory. Quarterly Journal of Experimental

Psychology, 1971, 23, 316-323. Craik, F. I. M., & Lockhart, R. S. Levels of processing: A framework for memory research.

Journal of Verbal Learning and Verbal Behavior, 1972, 11, 671-684. Fajnsztejn-Pollack, G. A. A developmental study ofdecay rate in long-term memory. Journal

of Experimental Child Psychology, 1973, 16, 225-235. Flavell, J. H. Developmental studies of mediated memory. In H. W. Reese & L. P. Lipsett

(Eds.), Advances in child development and behavior. New York: Academic Press, 1970. Vol. 5.

Flavell, J. H., & Wellman, H. Metamemory. In R. V. Kail& J. W. Hagen (Eds.), Perspectives on the development of mem0r.y and cognition. Hillsdale, N.J.: L. Erlbaum, 1977.

Gibson, E. J. Principles of perceptual learning and development. New York: Appleton- Century-Crofts, 1969.

Hagen, J. W. Some thoughts on how children learn to remember. Human Development, 1971, 14, 262-271.

MEMORIES OF MOROCCO 27

Hagen, J. W., Jongeward, R. H., & Kail, R. V., Jr. Cognitive perspectives on the development of memory. In H. W. Reese (Ed.), Advances in child development and behavior. New York: Academic Press, 1975. Vol. 10.

Hagen, J. W., Meacham, J. A., & Mesibov, G. Verbal labeling, rehearsal, and short-term memory. Cognitive Psychology, 1970, 1, 47-58.

Hardy, G., & Brunot, L. L’enfant marocain. Rabat, Morocco/Paris: Editions du Bulletin de L’enseignement du Maroc, No. 63, 1925.

Jenkins, J. J. Remember that old theory of memory? Well, forget it! American Psychologist, 1974, 29, 785-795.

Kagan, J., & Klein, R. E. Cross-cultural perspectives on early development. American Psychologist, 1973, 28, 947-961.

Kagan, J., Klein, R. E., Haith, M. M., & Morrison, F. J. Memory and meaning in two cultures. Child Development, 1973, 44, 221-223.

Keeney, T. J., Cannizzo, S. R., & Flavell, J. H. Spontaneous and induced verbal rehearsal in a recall task. Child Development, 1967, 38, 953-966.

Kingsley, P. R., & Hagen, J. W. Induced versus spontaneous rehearsal in short-term memory in nursery school children. Developmental Psychology, 1969, 1, 40-46.

Kintsch, W. Models for free recall and recognition. In D. A. Norman (Ed.), Models of memory. New York: Academic Press, 1970.

Levy-Bruhl, C. How natives think. New York: Washington Square Press, 1966, (originally published, 1910).

Liben, L. S., & Drury, A. M. Short-term memory in deaf and hearing children in relation to stimulus characteristics. Journal of Experimental Child Psychology, 1977, 24, 60-73.

Meacham, J. A. The development of memory abilities in the individual and society. Human Development, 1972, 14, 205-228.

Morrison, D. F. Multivariate sfatistical methods. New York: McGraw-Hill, 1967. Morrison, F. J., Holmes, D. L., & Haith, M. M. A developmental study of the effect of

familiarity on short-term memory. Journal ofExperimental Child Psychology, 1974, 18, 412-425.

Neisser, U. Cognitive psychology. New York: Appleton-Century-Crofts, 1967. Nelson, K. E. Memory development in children: Evidence from nonverbal tasks.

Psychonomic Science, 1971, 25, 346-348. Olson, G. M. Developmental changes in memory and the acquisition of language. In T. E.

Moore (Ed.), Cognitive development and the acquisition of language. New York: Academic Press, 1973.

Piaget, J. On the development ofmemory and identity. Worcester, Mass.: Clark University Press, 1968.

Porteus, S. D. Primitive intelligence and environment. New York: Macmillan, 1937. Scribner, S. Developmental aspects ofcategorized recall in a West African society. Cognitive

Psychology, 1974, 6, 475-494. Shepard, R. N., & Teghtsoonian, M. Retention of information under conditions approaching a

steady state. Journal of Experimental Psychology, 1961, 62, 55-59. Simon, H. A. How big is a chunk? Science, 1974, 183, 482-488. Tulving, E. Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.),

Organization of memory. New York: Academic Press, 1972. Wagner, D. A. The development of short-term and incidental memory: A cross-cultural study.

Child Development, 1974, 45, 389-396. Wagner, D. A. The effects of verbal labeling on short-term and incidental memory: A

cross-cultural and developmental study. Memory & Cognition, 1975, 3, 595-598. Wagner, D. A. Ontogeny of the Ponzo illusion: Effects of age, schooling, and environment.

International Journal of Psychology, 1977, 12 (in press). Wellman, H., Ritter, A., & Flavell, J. H. Deliberate memory behavior in the delay reactions of

very young children. Developmental Psychology, 1975, 11, 780-787.

28 DANIEL A. WAGNER

Wickelgren, W. A. Age and storage dynamics in continuous recognition memory. Developmental Psychology, 1975, 11, 165-169.

Witkin, H. A. Cognitive style approach to cross-cultural research. InternationalJournal of Psychology, 1967, 2, 233-250.

Yates, F. A. The art of memory. Chicago: University of Chicago Press, 1966.

REFERENCE NOTES

1. Lancy, D. F. Studies of memory in culture. Paper presented at the Conference on Issues in Cross-cultural Research, New York Academy of Science, 1975.

2. Sharp, D., Cole, M., & Lave, C. Education and cognitive development: The evidence from experimental research. Unpublished manuscript. Rockefeller University, New York, 1976.

3. Smith, J. E. K. Some simpliJed indicices ofsensitivity and bias. Michigan Mathematical Psychology Program, Report No. 74-5, University of Michigan, Ann Arbor, 1974.

4. Wagner, D. A. Memories of Morocco: A cross-cultural study of the influence of age, schooling and environment on memory. Ph.D. Dissertation, University of Michigan, Ann Arbor, 1976.

(Accepted September 19, 1977)