Flavell 1970 Dpmental Changes
Flavell 1970 Dpmental Changes
Flavell 1970 Dpmental Changes
JOHN H. FLAVELL
University of Minnesota
ANN G. FRIEDRICHS
University of North Carolina
JANE D. HOYT
University of Minnesota
Recent evidence suggests that the cognitive and linguistic skills which
the child spontaneously applies to the solution of memory tasks may show
dramatic changes with age, especially during the early school years
1 This research was supported in part by a research grant to the first author from the
National Institute of Child Health and Human Development (HD 0 1888) and by grants to
the University of Minnesota’s Center for Research in Human Learning from the National
Science Foundation (GS 541), from the National Institute of Child Health and Human
Development (HD 01136), and from the Graduate School of the University. The authors
are particularly grateful to Mervyn 0. Bergman for his technical assistance. They also thank
Mr. John H&man, principal of Lexington Elementary School, St. Paul, Minnesota, and
Mrs. Charles Pearlman, director of the Adath Jeshurun Nursery School, Minneapolis, Min-
nesota, for their cooperation in providing Ss.
324
MEMORIZATION PROCESSES 325
(Flavell, 1970). For example, older children are much more likely than
younger ones to use verbal rehearsal as a deliberate mnemonic technique
in serial recall tasks (Flavell, Beach, & Chinsky, 1966; Kingsley &
Hagen, 1969). There is now reason to suspect, however, that even such
relatively simple and homogeneous-looking memorization strategies as
verbal rehearsal may prove on closer analysis to be divisible into distinct
subvarieties, each one possibly acquired at a different point in the child’s
development (Daehler, Horowitz, Wynns, & Flavell, 1969; Flavell,
1970).
The present study was designed to obtain very detailed records of
children’s spontaneous verbal and visual-perceptual activities as they
memorized the names of a series of depicted objects. The purpose was to
identify whatever different study patterns of looking and verbalization
might be found, and to plot any developmental changes these patterns
might exhibit. A second objective was to discover something about possi-
ble developmental changes, not in the child’s favored strategies for memo-
rizing, but in the knowledge he has concerning his own memory. We are
aware of no previous research on the development of this sort of knowl-
edge, although it is currently attracting the attention of students of adult
memory processes (Tulving & Madigan, 1970).
Two procedures were used. One attempted to assess the child’s ability
to predict his own immediate memory span. The other estimated the
child’s ability to sense when a set of items he had been studying was suf-
ficiently well memorized to guarantee perfect immediate recall. Both of
these procedures have to do with the child’s awareness of his own storage
and retrieval capabilities, but in somewhat different ways. The first deals
with self-estimated, projected capacity to store and retrieve, the child
having in mind a future opportunity to study sets of items of a certain
type. The second deals with his self-estimated ability to retrieve here and
now, while in process of studying and storing a specific set of items.
METHOD
Subjects
The Ss were 14 nursery school children, 28 kindergarteners, 28 second
graders, and 14 fourth graders from suburban, largely middle-class homes,
with equal numbers of boys and girls within each grade level.
Procedure
All testing was administered by a female E (the second author) in a two-
room trailer-laboratory parked adjacent to S’s school. The testing
sequence, identical for all Ss, is given below:
326 FLAVELL, FRIEDRICHS, HOYT
Predicted object span. The child’s first task was to estimate his own
immediate memory span. The E had at her disposal strips of heavy paper
on which were mounted different-length series of pictures of familiar,
readily namable objects (e.g., toy blocks, scissors, house). She first
showed S a strip containing one object. She then named it, covered it, and
asked him if he thought he would now be able to say its name from mem-
ory. He could, of course, and was asked to say the name aloud if he had
not already done so spontaneously. The same procedure was then repeated
for a horizontally presented strip of two pictures, with E emphasizing the
need for ordered, left-to-right recall of the object names. The prediction
process continued (now, minus actual attempts at recall) for a strip of
three pictures, then four, and so on, either until S estimated that the series
had now become too long for him to recall, or until a series of 10 pictures
had been presented. The E’s instructional formula for each series of three
pictures or more was: “Do you think you could remember n words (points
to a strip of n pictures) if I said each of them just once and then covered
them up?” An S’s predicted object span was taken to be the longest series
of object names (to a maximum of 10) he thought he would be able to
recall correctly.
Actual object span. The next step was to assess S’s actual memory span
for such items. The E read aloud a series of first three, then four, etc.,
familiar object names at the rate of one name per second, with S in-
structed to try to repeat each series in correct order. If S failed a
sequence, he was given one more try with another sequence of the same
length. In analogy with predicted object span, an S’s actual object span
was defined as the longest series he was actually able to recall correctly.
Memorization and recall. The S was next asked to name the object
depicted in each of 33 small, black and white line drawings, with E provid-
ing help when necessary. The objects were commonplace ones, likely to
be familiar to young children, and the pronunciation of their names in-
volved fairly conspicuous and distinctive lip movements (e.g., ball, moon,
cat). All the object pictures that S would subsequently be called upon to
study and memorize, in the course of one practice trial and three test
trials, were drawn from this set of 33.
The S was then seated before a stimulus panel. The panel was mounted
in the one-way mirror of the wall which separated the two rooms of the
trailer-laboratory. The panel contained a horizontal row of ten 2 X 3-in.
windows with a button directly beneath each window. One of the above-
mentioned object pictures was illuminated in a window whenever S
pressed the button beneath it, and it remained visible only so long as S
kept pressing the button. On the initial practice trial, only the rightmost
three windows contained pictures. The instructions to each S on this trial
were approximately as follows:
MEMORIZATION PROCESSES 327
“Now the pictures (referring to the ones S had just named) are going to
be hidden. You will have to push a button like this to see them (E pushes
the three buttons in a random sequence). Now you try it, but only push one
button at a time. You can look at any one you need to and in any order
you want, but I’m going to want you to say them back to me in this order
(gestures in the left-to-right direction)-first this one, then this one, then
this one (points to each window), all without looking at the pictures. Let’s
try it. Remember, you can look at the pictures as long as you want. When
you are sure you know them all by heart, ring this bell (a desk bell is on
the chair beside S) and I’ll come hear you say them back to me. Be
careful, though, not to ring until you can remember every one very, very
well in the right order without looking. So when you do know them, ring
the bell and say the words right out loud like this (E demonstrates with the
three pretest pictures), and I’ll hear you. Okay, you can begin.” If, during
this practice trial, S made procedural errors, they were corrected (e.g., he
was not permitted to depress more than one button at a time). If he failed
to recall all the words in correct order, he was cautioned once again not to
ring the bell until he is sure he could remember the whole series. If he per-
formed successfully, he was warmly praised.
Three test trials followed. The procedure was essentially identical to
that of the practice trial, with two exceptions: the object pictures were
now placed behind the leftmost rather than rightmost windows of the
panel; the above-mentioned cautioning after imperfect recall was discon-
tinued after the first of these three trials. A different series of objects was
presented on each trial, but all three series were set exactly equal in length
to the individual S’s own actual object span, determined minutes earlier in
the testing session. Pilot research had suggested that a series of object-
span length would be of suitable difficulty level for our purposes-defi-
nitely short enough to be able to be recalled by that child, but only after
some reasonable effort at memorization on his part. The particular object
pictures presented for study on each trial always consisted of the first II
pictures of a fixed sequence of 10, where n equals that S’s actual object
span. These three lists of 10, plus the three practice items, comprise the
33 object pictures S had named initially. The rare child who on any trial
persisted in waiting as long as 15 min before indicating his readiness to be
tested was dropped as an S; the average length of study period for all Ss
retained in the sample was actually only 53.43 sec.
Recall for each series was also assessed a second time with no explicit
prior announcement to S that this would occur. These retests were in-
serted into the study-test sequences as follows: study 1 (i.e., self-timed
study of the first picture series)-test 1 (i.e., the immediately-subsequent
attempt to recite this series to @--study 2-test 2-retest l-study 3-
test 3-retest 2-retest 3. The retests were presented to the child as
328 FLAVELL, FRIEDRICHS, HOYT
minor excursions from the main procedure (which, indeed, they were to
us as well), without any suggestion conveyed that S really should be ex-
pected to perform well on them.
Finally, half of the kindergarten and second-grade groups and all of the
preschoolers and fourth graders did their studying with E physically ab-
sent from the room (she awaited S’s ready signal just outside the closed
door). In the case of the remaining kindergarteners and second graders, E
remained in the room, sitting quietly a few feet behind him and unrecep-
tive to any bids for attention while he studied.
Rationale. Each part of the foregoing task sequence was designed to
provide information related in some way to the development of memory.
It is apparent, for instance, that the relation between the child’s predicted
memory span and his actual memory span might inform us about one as-
pect of his knowledge regarding his own memory capacities. His ability to
recall the picture series on the three study-test trials should be a measure
of another aspect of that knowledge, namely, knowledge about the re-
trievability of currently processed items. It seems reasonable to credit a
child with at least a modicum of this latter sort of knowledge if, having
said he is now ready to recall a series perfectly, he then proceeds to do
just that.
The retesting of previously tested series was something of an after-
thought, inserted into the design in the hope that it might at least roughly
measure degree of overlearning or “overmemorization” of correctly
recalled series. It might be, for example, that there is an increase with age
in the ability or disposition to memorize the series just well enough to sat-
isfy E’s immediate recall requirements; the use of this efficient, “least-ef-
fort” strategy could be reflected in perfect recall at test coupled with very
poor recall at subsequent retest.
The purpose of the self-timed study periods was to permit S to memo-
rize the picture series and, more importantly, to afford us some observa-
tional data as to how he went about doing so, i.e., his study-memorization
strategies. An Esterline-Angus event recorder was connected to the but-
tons of the display panel, its pen tracings automatically providing a per-
manent record of the timing and sequence of S’s spontaneous picture-ex-
posing behavior (and hence, by not much inference, of his spontaneous
picture-studying activity). An observer (the third author) was seated
directly behind the one-way mirror throughout each testing session, She
had trained herself to lipread the particular object names used in this
study, was always aware of which object picture the child was currently
studying, and could observe his face from a distance of l-2 ft. In addition,
she wore a set of high-quality earphones to help supplement the visual
evidence of verbalization. Pilot research had indicated the desirability of
329
MEMORIZATION PROCESSES
We shall begin with an analysis of the data most directly pertinent to the
child’s developing knowledge of his own memory abilities and processes.
These data consist, firstly, of the relation between S’s predicted and ac-
tual object span, and secondly, of his readiness to recall items after the
three self-terminated study periods. The remainder of the paper describes
and interprets various behavior patterns observed during the study
periods themselves, and also draws some general conclusions regarding
the nature and development of memorization processes.
Predicted versus Actual Object Span
TABLE 1
Performance on Actual Object Span and Predicted Object Span Tests
Recall Readiness
Table 2 shows the percentages of Ss at each grade level who proved
themselves ready to recall sets of picture series perfectly on none, one,
two, or all three of their post-study-period recall opportunities. It is appar-
ent that there is a very marked improvement over this age range in the
child’s ability to sense when he has memorized a set of items sufficiently
well to recall them perfectly. Combining adjacent grades and adjacent
levels of recall to make a fourfold contingency table, this improvement is
highly significant (x2 = 26.41; df= 1; p < .OOl). One might have ex-
pected that there would be some sort of grade by trial interaction here,
with perhaps the younger Ss, having profited from knowledge of their ini-
tial recall failure and from E’s cautioning to do better next time, improving
on the later trials. No interaction of this sort was evident in the data, how-
ever. Again, one certainly might have predicted that “realistic” (< 10) ob-
ject span predictors would also tend to be more recall-ready, i.e., to recall
better at test. The object span prediction test and the recall tests were,
after all, both designed to tap S’s knowledge or skill with respect to his
own memory processes. Again, however, the data suggested no such rela-
tionship. And finally, the retesting of previously tested picture series also
yielded nothing of interest: even when initial recall of a series had been er-
TABLE 2
Recall Readiness: Percentages of Ss Performing
Without Error on O-3 Recall Tests
None 14 18 0 0
One 64 32 7 0
Two 14 36 11 14
Three 7 14 82 86
332 FLAVELL, FRIEDRICHS, HOYT
rorless, subsequent recall of the same series tended to be very poor, and
more or less uniformly so across grade levels. These negative findings
notwithstanding, the object span prediction and recall readiness data
impress us as being decidely encouraging overall. Even measures as crude
and insensitive-appearing as these have proved capable of suggesting, and
rather strongly, that S’s cognitive relation to his own memory system does
somehow change in the course of the early school years. But more probing
and analytic studies will now be needed, obviously, to find out exh,tly
what the “somehow” consists of.
grade group. It is possible that this increase largely reflects the older Ss’
greater effort to insure recall readiness-an effort that, as we have just
seen (Table 2), was decidedly successful. Length of study period also
varied significantly fp < .05) with trial, the third study period tending to
be shorter than the first and second. There was no significant grade by
trial interaction here, however.
The question of principal interest to us, however, concerned the possi-
ble influence of several variables, and most especially the age variable, on
the four study patterns. Four-way analyses of variance were performed
upon the study pattern scores of the 56 E-out Ss (the 28 E-in Ss were not
considered in these analyses, in order to maintain equal Ns in the four age
groups). The four independent variables were grade (4 levels), sex (2),
trial (3), and “trial quartile” (4), i.e., each self-timed study period was
divided into four equal time segments. The first two are between-S vari-
ables and the latter two are within-S variables. As Table 3 shows, there
were significant grade, trial quartile, and grade x trial quartile effects in
the case of all four study period behaviors. There were only three other
significant effects out of the 48 remaining possibilities, none of them very
credible, interpretable, or interesting: (1) a sex X trial quartile interaction
0, < .Ol) for Naming (both sexes show what will be seen to be a charac-
teristic distribution of Naming responses across the four quartiles of each
trial, but girls show it a bit more strikingly than boys); (2) a grade X sex X
trial interaction (p < .05) for Anticipation, essentially indescribable; (3) a
main effect for trial 01 < .05) in the case of Rehearsal that is certainly
puzzling if it is not a chance effect (more rehearsal responses in the sec-
ond trial than in the first or the third).
In contrast with these rather suspect effects, the pattern of findings
represented in Table 3 tells what seems to us to be a rather clear and inter-
esting story. Figure 1 gives a visual representation of each of the four
TABLE 3
F Values and Associated p Values for Significant Analyses of Variance
Effects Involving Study Period Behaviors
Variable df F P F P F P F P
Grade (G) 3,48 4.09 c.05 9.88 c.001 9.38 <.OOl 7.41 <.OOl
Trial quartile (Q) 3,144 27.08 <.OOl 9.36 < .OOl 22.80 <.OOl 9.06 c.001
GxQ 9,144 7.55 <.ool 2.39 C.05 3.24 <.Ol 5.60 <.OOl
MEMORIZATION PROCESSES 335
I I /
M Grade N _
0-0 Grada h
0-0 Grade 2
A-A Grade 4
pattern, summing across the three trials: more instances of Naming in the
first two trial quartiles combined than in the last two combined, and the
converse for Anticipation and for Rehearsal.
What functions might Anticipation and Rehearsal perform in S’s at-
tempts at memorization, and why should they gradually preempt Naming
as the study period continues? Our belief is that, relative to Naming,
Anticipation and Rehearsal have somewhat more to do with the output
from memory, or retrieval, in contrast to the input to memory, or storage.
And it is item retrieval rather than item storage, after all, that E will
demand of S when the study period is completed. Our hypothesis is that,
as the study period proceeds, the process of getting ready to be tested
becomes increasingly one of checking and maintaining one’s gradually
improving ability to retrieve a complete, correctly ordered list of the pic-
ture names. While it is undoubtedly true that the exercise of all three
study patterns can and does serve to improve retrievability, Anticipation
and Rehearsal seem especially well-fitted to monitor and maintain it as
well. Like Naming, Anticipation and Rehearsal do of course consist of
potentially recall-facilitating repetitions of item names. Unlike Naming,
however, an attempted Anticipation also provides information about an
item’s current level of retrievability. It thus serves to check and monitor
S’s progress towards the task-defined criterion of complete retrievability.
Also unlike Naming, Rehearsal serves to sustain and safeguard the list’s
gradually improving retrievability, the rehearsed utterance eventually
becoming indistinguishable from the recall-test utterance.2 Needless to
say, Anticipation could also maintain as well as monitor, and Rehearsal
automatically monitors in the very act of maintaining. We saw a number
of rather amusing examples of Rehearsal apparently serving both roles.
An S (usually an older one), finally able to rehearse his entire list (call it
“A-B-C”) rapidly and correctly, would thereupon wrap up both study
period and recall test as follows: “A-B-C, ” “A-B-C” (hits the signal bell
without the slightest pause in the rehearsal cycle) “A-B-C,” “A-B-C” (E
hastens into the room), “A-B-C” (S turns to E), “A-B-C!” (said loudly
and triumphantly to E). By virtue of having produced one or more
complete rehearsals, such an S judges that he is now ready to be tested,
i.e., he thinks he can produce yet another, identical rehearsal for E that
will comprise his recall test (a monitoring function). And by virtue of then
2 This last would of course only be true of the more common form of Rehearsal, wherein S
repeats a whole string of object names. The less common form, i.e., repeating a single name
one or more times, strikes one as a rather different sort of process altogether-more like an
elaborated version of Naming, perhaps. It was unfortunate that the observer could not
quickly, easily, or frequently enough distinguish the two forms to warrant recording them
under separate response categories.
MEMORIZATION PROCESSES 337
REFERENCES
DAEHLER, M. W., HOROWITZ, A. B., WYNNS, F. C., & FLAVELL, J. H. Verbal and non-
verbal rehearsal in children’s recall. Child Development, 1969, 40, 443-452.
DALE, P. S. Color naming, matching, and recognition by preschoolers. Child Development,
1969,40, 1135-l 144.
FLAVELL, J. H. Developmental studies of mediated memory. In H. W. Reese & L. P. Lips&t
(Eds.), Advances in child development and behavior. Vol. 5, New York: Academic
Press, 1970.
FLAVELL, J. H., BEACH, D. H., & CHINSKY, J. M. Spontaneous verbal rehearsal in a
memory task as a function of age. Child Development, 1966, 37, 283-299.
340 FLAVELL, FRIEDRICHS, HOYT
HAGEN, J. W., MEACHAM, J. A., & MESIBOV, G. Verbal labeling, rehearsal, and short-term
memory. Cognitive Psychology, 1970,1, 47-58.
KINGSLEY, P. R., & HAGEN, J. W. Induced versus spontaneous rehearsal in short-term
memory in nursery school children. Developmental Psychology, 1969, 1, 40-46.
KOHLBERG, L., YAEGER, J., & HJERTHOLM, E. Private speech: four studies and a review of
theories. Child Development, 1968, 39,691-736.
TULVING, E., & MADIGAN, S. A. Memory and verbal learning. Annual Review of Psychol-
ogy, 1970,21,437-484.