Content uploaded by Mark A. Mcdaniel
Author content
All content in this area was uploaded by Mark A. Mcdaniel on Feb 16, 2016
Content may be subject to copyright.
Journal of Expcrimental Psychology:
Learning, Memory, and Cognition
1987,
Vol. 13, No. 2, 291-300
Copyright 1987 by Ihc American Psychological Association, Inc.
0278-7393/87/SOO.75
Generation and Precision of Elaboration: Effects on
Intentional and Incidental Learning
Michael Pressley
University of Western Ontario, London, Ontario, Canada
Mark
A.
McDaniel
University of Notre DameJames
E.
Turnure
University of Minnesota
Eileen Wood and Maheen Ahmad
University of Western Ontario, London, Ontario, Canada
Across three experiments adults were presented either base sentences containing arbitrary relations
(e.g., The fat man
read the sign)
or base sentences accompanied by a precise elaboration that made
clear the significance of the particular type of
man
doing the action
{e.g.,
The fat man
read the sign
warning about
thin
ice).
Subjects either read the sentence alone or answered a question accompany-
ing the sentence. Why did
that particular
man do
that?
accompanied base sentences;
How does
the
last part
of
the sentence
make
clear why that particular
man did (hat? accompanied precisely elabo-
rated
sentences.
Both intentional and incidental learning
were
studied. The
positive
effects of provid-
ing precise elaborations were moderate in size and confined to incidental learning. Much larger
incidental and intentional learning gains followed generation of elaborations in response to ques-
tions,
with generated precise elaborations facilitating acquisition slightly better than generated im-
precise elaborations.
Much contemporary cognitive research
is
directed at analysis
of learning from written texts and verbal discourse
(e.g.,
Brans-
ford et
al.,
1982;
van Dijk
&
Kintsch,
1983).
For instance, Stein,
Bransford, and their associates (Stein
&
Bransford,
1979;
Stein,
Littlefield, Bransford,
&
Persampieri, 1984) reported a series of
experiments in which adult subjects were presented with text
that contained (a) arbitrary relations with intersentence confu-
sions likely (base sentences),
(b)
elaborations that made the oth-
erwise arbitrary relations more understandable (precise elabo-
rations), or
(c)
elaborations that
were
consistent with the mean-
ing of the sentence but did not clarify the significance of the
relation specified in the sentence (imprecise elaborations). The
following
is
a subset of Stein's and Bransford's base sentences:
The
hungry
man got into the
car.
The strong man
helped
the
woman.
The brave man
ran
into the
house.
These arbitrary sentences can be expanded, however, with pre-
cise elaborations:
The
hungry
man got into the car to go to the
restaurant.
This research
was
funded
by a
grant
to the
first
author from the Natu-
ral Sciences and Engineering Research Council of Canada Grant
T1S5A2.
Special
thanks to
Albert
Katz,
Allan U.
Paivio,
John Bransford, John
Gardiner, Henry L. Roediger III, and an anonymous reviewer for con-
structive critiques of a previous draft of this article.
Correspondence concerning this article should be addressed to Mi-
chael Pressley, Department of Psychology, University of Western On-
tario,
London, Ontario, Canada N6A 5C2.
The strong man helped the woman carry the heavy packages.
The brave man ran into the house to
save
the baby from the fire.
Alternatively, the sentences could be elaborated imprecisely:
The hungry man got into the car to
go
for a ride.
The strong man helped the woman find the newspaper stand.
The brave man ran into the house to
find
some ketchup.
Stein and Bransford argue that this work is significant be-
cause acquisition of information from arbitrary sentences mir-
rors acquisition of information by novices. Novices, lacking
topical knowledge, often do not automatically link newly en-
countered relations to information that could make the signifi-
cance of the relations more understandable and thus, make the
material more memorable (e.g., Bransford et al., 1982). Stein
and Bransford and their associates further propose that learning
can be enhanced by helping the learner to reduce the apparent
arbitrariness of the to-be-learned materials (e.g., Bransford et
al.,
1982;
Stein et al.,
1984).
The precisely elaborated sentences
were designed to test this hypothesis. Consistent with Stein and
Bransford's theoretical analysis of their materials, adults inci-
dentally learn precisely elaborated text better than base sen-
tences or imprecisely elaborated sentences. In particular, adult
subjects can answer more who questions
(e.g.,
Who got into the
car?)
on an unexpected test following presentation of precisely
elaborated sentences than following presentation of base or im-
precisely elaborated sentences. This
finding
is subsequently re-
ferred to (following Levin, 1976) as the imposed precise-elabo-
ration effect.
The main purpose of the experiments reported here was to
determine if it is possible to improve learning beyond gains as-
sociated with provision of precise elaborations. In particular,
291
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
292PRESSLEY, McDANIEL, TURNURE, WOOD, AHMAD
we hypothesized that self-generated elaborations that
are
prompted by questions about the base sentences might be more
potent than providing
precise
elaborations
to
learners.
This
self-
generated elaboration hypothesis
was
based on
a
number of
ob-
servations from the literature.
In general, verbal material that
is
produced
by a
learner
is
remembered better than similar material that
is
presented
to
the learner (Jacoby, 1978; McDaniel, 1984; Slamecka &
Graf,
1978).
For
instance, sentences that
are
produced
by
learners
are remembered better than sentences that
are
provided
by an
experimenter (Glover, Bruning,
&
Plake,
1982; Graf, 1980,
1982).
When Stein
and
Bransford (1979, Experiment 1) asked
their adult subjects to construct continuations
for
the base sen-
tences used
in
that study, recall
of
the
sentences that were
pre-
cisely elaborated
was
over 90% compared with 49% recall
for
sentences elaborated imprecisely and
74%
recall
in
an imposed
precise-elaboration condition. Hashtroudi, Parker, DeLisi,
and
Wyatt (1983) replicated the Stein and Bransford (1979, Experi-
ment
1)
sentence continuation
and
imposed precise-elabora-
tion procedures
and
obtained
the
same pattern
of
results
re-
ported
by
Stein
and
Bransford
in
those conditions. Although
this pattern
of
outcomes suggests that self-generated precise
elaborations
are
more potent than imposed precise elabora-
tions,
the
items that were precisely elaborated
in
those experi-
ments probably suggested such elaborations more than did
the
sentences that were elaborated imprecisely (i.e.,
the
precisely
elaborated sentences were self-selected).
It
could
be
that
sen-
tences suggesting precise elaborations
may
also
be
inherently
more memorable
for
some reason. Thus,
in a
second experi-
ment Stein and Bransford used
a
different tactic to study gener-
ated elaborations. Stein
and
Bransford (1979, Experiment
2)
explicitly compared learners
who
were instructed
to
generate
precise elaborations
(by
answering
Why
questions about
the
sentences) with learners
who
were provided precise elabora-
tions.
In
that experiment Stein
and
Bransford found
no mne-
monic facilitation for generated
elaborations.
A
potential expla-
nation
for
the lack of effect
in
that study, however, was that
the
generated and imposed precise-elaboration means
were
close
to
the ceiling.
Following Stein and Bransford (1979, Experiment
2),
the ap-
proach that
we
elected
for
eliciting subject-generated precise
elaborations was
to
have learners answer why questions about
the relations specified in base sentences
(e.g.,
Buium
&
Turnure,
1977;
Turnure, Buium,
&
Thurlow, 1976). Thus, there were
conditions
in
which subjects answered
the
question, Why did
that particular
man do that? with instructions emphasizing the
production of
a
precise elaboration linking
the
man's character-
istics
to
what
he did.
Extensive pilot testing
was
conducted prior
to these experiments to determine
a
list length that would pro-
vide data
far
removed from the performance ceiling.
A second goal of these experiments
was
motivated by the fact
that
all
published studies
of
imposed precise elaboration with
adults have been limited
to
occasions when learning was inci-
dental.1
This seems
an important limitation given that the work
on precise elaboration is considered important
for
understand-
ing how people acquire and effectively use knowledge
in
formal
educational contexts (e.g., Bransford
&
Stein, 1984), where
learning
is
often intentional. It
may be
that
the
imposed precise-
elaboration effect
is
substantially reduced
or
eliminated when
adult learning
is
intentional compared
to
when
it is
incidental.
In fact,
it
could
be
that
the
intentional learning
of
base
sen-
tences might even exceed
the
intentional learning
of
sentences
that include precise elaborations. These possibilities seem plau-
sible given that
(a)
adults
are
often spontaneously elaborative
when given intentional learning instructions (e.g., McDaniel
&
Kearney, 1984; Pressley, 1982; Rohwer, 1973)
and
(b) self-gen-
erated elaborations may be more potent than imposed elabora-
tions as outlined
in
the foregoing discussion. Thus, Experiment
1 examined both intentional
and
incidental learning
in im-
posed precise-elaboration and base sentence conditions. Exper-
iments
2
and
3
replicated and extended respectively
the
inciden-
tal and intentional learning results of Experiment
1.
Experiment
1
To test
the
hypotheses outlined
in the
introduction, Experi-
ment 1 examined incidental
and
intentional learning
of
arbi-
trary relations contained in sentences, with sentences presented
in three different ways. Subjects studied either base sentences,
base sentences accompanied by the question, Why did that par-
ticular man
do
that?,
or precisely elaborated versions of the base
sentences.
Method
Subjects.
A
total
of
120
undergraduates
(83
female,
37
male) enrolled
in introductory psychology
at a
large Canadian university participated
as a course option. Subjects ranged
in
age from
17
to
44 years
(M
=
19
years
8
months,
SD
=
3
years
1
months). Twenty subjects
were
randomly
assigned
to
each
of
the
six conditions produced by the factorial combi-
nation
of
the intentional learning variable (intentional vs. incidental)
and
the
acquisition context variable (base, subject-generated elabora-
tion
in
response to the
Why
question,
imposed precise elaboration).
Materials.
Three sets
of 24
stimulus
sentence
cards were
constructed.
The sentences
on
each card were typed
in
capital letters
on
white
12.7
cm
X
20.3 cm
(5
in.
X
8
in.)
cards,
one stimulus to a card.
One of the sets (used
in
the two base conditions) contained base sen-
tences
in
which particular
men
performed actions.
The
first
10 sen-
tences
in the set
were used
in
Stein
et al.
(1984).
The
remaining
14
sentences were constructed
by
randomly assigning 14 types
of
men
to
14
sentence predicates, with
five
judges agreeing independently that the
resultant sentences were arbitrary
as
denned by Bransford et
al.
(1982).
The entire base sentence was underlined
on the
presentation card.
See
the Appendix for
the
24 base sentences
in
their order of presentation.
The second
set of
24
cards (used
in the
imposed precise-elaboration
conditions) contained precisely elaborated versions
of the
base
sen-
tences with only
the
base portion
of
the
sentence underlined.
The
first
10
elaborations were used by Stein
et
al. (1984); the remaining
14
were
judged
to
be precise elaborations by
the
five
judges. These elaborations
are
also
presented in the Appendix.
The third
set of
cards (used
in the
questioned conditions) included
the
24
base sentences, each underlined, with
the
question, "WHY DID
THAT
PARTICULAR
MAN DO THAT?"
printed
below
the
base
sentence.
1
All
of the previously published intentional learning work in this par-
adigm has involved children. These studies have focused
on the
types
of elaborations that children of different abilities make as they process
base sentences,
on
whether children
can
be trained
to
generate precise
elaborations
in
response
to
base sentences,
and on
establishing charac-
teristics
of child
learners
who
benefit from imposed
precise
elaborations
(Franks etal.,
1982;
Stein etal.,
1982;
Wong
&Sawatsky, 1984).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
PRECISE ELABORATIONS293
Three sets of three practice sentences, typed on individual cards in
the same format of the to-be-learned stimuli, were used when instruc-
tions were given to subjects. Subjects in the base conditions were pre-
sented with the following:
THE UGLY MAN BOUGHT SOME PLASTIC.
THE
TOOTHLESS
MAN WROTE
A
CHECK.
THE STRONG MAN CARRIED
A
SHOVEL.
Subjects in the questioned conditions were presented the same sen-
tences with the question,
"WHY DID THAT PARTICULAR MAN
DO
THAT?"
printed below each sentence. Subjects in the imposed precise
elaboration condition
saw
the following elaborated sentences:
THE UGLY MAN BOUGHT SOME PLASTIC TO MAKE A MASK.
THE TOOTHLESS MAN WROTE A CHECK TO PAY FOR NEW DEN-
TURES.
THE STRONG MAN CARRIED
A
SHOVEL TO DIG OUT HEAVY ROCKS.
The memory test was composed of
24 Who
questions, one for each
sentence. Each question required recall of the type of man doing the
action specified in the base sentence (e.g., Who gave money to the
robber?).
All responses made
by
subjects were audio recorded on a Sony
TCM-757 cassette recorder that was in full view of the subjects so that
they knew when they
were
being recorded.
Procedures.
Subjects in the incidental/base and incidental/imposed
precise-elaboration conditions were told that the purpose of the study
was to determine how easy certain sentences were to understand. They
were told that they would be shown one sentence at a time, each one
about a particular type of
man
with the task of rating the sentences on
a scale from 1 (very easy to comprehend) to 7 {very difficult to
comprehend).
They practiced doing this with the three practice sen-
tences and
were
told that they should say their ratings out loud.
Intentional/base and intentional/imposed precise elaboration sub-
jects
were
told that
the
purpose of the experiment
was to
determine how
well they could learn sentences. They
were
told that they would be see-
ing sentences at a 7.5-s rate and that they would have to remember the
sentences later. They practiced studying with the three sample items
followed
by a
test
on
those
items.
These
questions were
presented orally
by the experimenter: "Who carried the shovel? Who bought
some
plas-
tic? Who wrote a check?" (Practice questions were presented in a ran-
dom order differing from subject to subject.)
Incidental/questioned subjects were told that the purpose of the ex-
periment was to find out how well they could answer questions. They
practiced answering questions out loud using the three practice sen-
tences, each followed by,
"WHY DID THAT PARTICULAR MAN
DO
THAT?"
It was reiterated four times during the instructions that it was
very important to explain the significance of that particular type of man
doing the particular
action.
Subjects
were
told to answer out loud.
Intentional/questioned subjects received the same instructions as in-
cidental/questioned subjects except that
they were
told that
the
purpose
of the study was to see how well they could learn sentences and that
coming up with good answers was a means to learning the content of
the
base
sentences.
They practiced answering
the
three
sample test
ques-
tions.
In summary, incidental subjects did not know about
the
test
following
presentation of the sentences, whereas intentional subjects did. The in-
cidental task in the
base
and
the
imposed precise elaboration conditions
was
rating
comprehensibility of base and precisely elaborated sentences
respectively, whereas it was answering,
"WHY DID THAT PARTICULAR
MAN DO THAT?"
in the questioned condition. Intentional/base and in-
tentional/imposed precise-elaboration subjects were left to their own
devices to learn the
sentences.
Intentional/questioned subjects
were
in-
structed to use question answering as a means to remember the base
sentences.
The
sentences were
presented in
a
constant order, beginning with the
Table 1
Percentages
of
Questions Answered
on
Poststudy Test
as a
Function
of Experimental
Condition:
Experiment 1
Condition
Base
Base/questioned
Imposed precise-elaboration
Incidental
M
15.0
65.2
24.2
SD
11.3
21.3
13.4
Intentional
M
58.1
76.9
46.0
SD
22.8
13.4
17.9
Note.
Percentages based on 24 responses, n
=
20 per condition.
10 sentences from the Stein et al. (1984) research. The experimenter
showed each of the 24 sentences for 7.5 s per card with approximately
1 s between cards with subjects in all conditions informed of
this
rate
of pacing.
Immediately after the presentation of the 24 sentences, subjects were
asked the 24
Who
questions corresponding to the sentences, with each
subject presented the questions in a different random order. Subjects
were permitted up to
10 s
to respond orally to an item.
After subjects completed this cued-recall task, a posttest interview
about processing during study was conducted in the base and imposed
precise-elaboration
conditions.
Subjects were shown
six
of the sentences
they had studied. In each condition 5 subjects responded to Sentences
1,5,9,
13,
17,
and
21; 5
responded to 2, 6, 10, 14, I8t and
22; 5
to 3, 7,
11,15,
19, and
23;
and
5
to 4, 8, 12, 16, 20, and
24.
The experimenter
instructed participants to tell out loud exactly what they had been
thinking when working on that
sentence.
It was emphasized to the sub-
ject to report actual processing during study, not what the subject now
felt could or should have been done. Subjects
were
permitted as
long
as
they needed to respond to items, although answers seldom required
more than 30 s.
After responding to the six probed items, subjects in base conditions
then heard an explanation of
precise
elaboration using
two
of the sam-
ple
sentences.
Subjects heard
For The
strong
man
carried
the
shovel
you might have thought to
yourself,
why would a
strong
man carry a shovel and then come up
with a reason—Perhaps it was a very heavy shovel, one made of
lead. Here's another possibility—Maybe he
was
going to dig heavy
rocks.
I'll illustrate this strategy that I'm interested in for another
of
the
sample sentences, The
toothless
man
wrote
a
check.
Why
would it
be
important that it
was a
toothless
man?
Maybe
the
check
was
to pay for
new
dentures.
If
so,
it would make sense to mention
that it
was
a toothless man.
Base condition subjects were then asked if (during study) they ever
made up reasons to clarify why the particular man did the particular
action. Subjects
who
said they did then estimated how many items were
responded to in this
way.
Results and Discussion
The main analysis was of percent correct recall on the post-
test. The means for each condition are presented in Table 1.
Because there were 12 contrasts that were of interest a priori,
these recall data
were
analyzed with
12
nonorthogonal contrasts
using the Dunn-Bonferroni procedure, a technique that maxi-
mized power and permitted testing of the exact hypotheses that
were
of interest while maintaining
a
reasonable Type
I
error rate
(Kirk,
1982).
An experimentwise error rate of.
15 was
adopted,
which is equal to the omnibus error rate in the 3
X
2 analysis
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
294PRESSLEY, McDANIEL, TURNURE, WOOD, AHMAD
of variance (ANOVA) alternative for these data
(i.e.,
with
Type
I
rate < .05 per effect, overall rate
<,.
15
in that
setup).
Thus, with
12
comparisons, the per comparison error rate
<,
.0125, cutoff
t{\
14)
= 2.53 {MSe
=
296.60).
The 12 contrasts tested (a) for all pair-wise differences be-
tween acquisition context conditions (i.e., base, base/ques-
tioned, imposed precise elaboration) within the incidental and
intentional levels, (b) for incidental versus intentional differ-
ences within levels of
the
acquisition context factor, and (c) for
the three 2X2 interactions possible within the 3 X 2 setup.
Under incidental instructions, questioned subjects recalled
more than subjects given the base sentences, f(l 14) = 9.22, and
more than imposed precise-elaboration subjects, f(l 14) = 7.54.
The performance of the incidental/base and incidental/im-
posed precise-elaboration subjects did not differ significantly,
t(\ 14) = 1.68. Under intentional instructions, questioned sub-
jects again answered more questions correctly than subjects in
the base condition, J(l 14) = 3.44, and subjects in the imposed
precise-elaboration condition, /(114) = 5.66. Although there
was a trend toward superior performance in the intentional/
base condition versus the intentional/imposed precise-elabora-
tion condition, the difference was not significant, at the a
priori
level,
t( 114)
-
2.22,
p <
.03.
Intentional recall was significantly
greater than incidental recall when base sentences were given,
/(114) = 7.92, and with imposed precise elaborations, t{\ 14) =
4.02,
but not when subjects were questioned, ((114) = 2.14. Fi-
nally, two of the three 2x2 interactions were significant. The
apparent crossover interaction involving the intentional and in-
cidental versions of the imposed precise elaboration and base
conditions was significant, |f(l
14)|
= 2.76, and the questioned
versus base difference was larger with incidental instructions
than with intentional instructions, /(114) = 4.08. The size of
the questioned versus imposed precise elaboration contrast did
not differ as a function of the intentional/incidental manipula-
tion, |/(114)| = 1.32.
Experiment
1
was revealing about the two main issues that
motivated the study: (a) Answering Why questions improved
performance, with the effect larger when learning
was
incidental
than when it was intentional. Trying to come up with precise
elaborations produced large effects (i.e., all relevant compari-
sons were greater than a standard deviation), (b) There was a
significant interaction between type of sentence (either base or
imposed precise elaborations) and whether a test was expected
or not. When learning was intentional, recall of
base
sentences
exceeded recall of imposed precisely elaborated sentences (al-
beit not at the a priori significance level of
.0125),
with the or-
dering of means reversed under incidental instructions. Al-
though the incidental/imposed precise elaboration versus inci-
dental/base contrast was not significant in the conservative
design and analysis used here, the results that we obtained are
in fact consistent with earlier reports. The effect size for this
contrast
was
0.74
SD,
calculated relative to the within-cell vari-
ation in the incidental/base and incidental/imposed precise-
elaboration conditions
(i.e.,
relative to the within-cell variation
in the two types of conditions common to all of the relevant
experiments; Hedges
&
Olkin,
1985,
chap.
5).
This
is
within the
range of effect sizes produced in previous studies
(e.g.,
0.72 SD
in Stein et al., 1984, Experiment 2). The nonsignificant trend
here
was
due largely to less power in this experiment for detect-
ing moderate-sized effects (Cohen, 1977) than was the case in
previous research that involved fewer
cells,
fewer comparisons
and generally larger per comparison Type I error rates.
The mechanisms mediating both of
these
main outcomes of
Experiment
1
are more obvious in light of supplementary data
collected during study and after the test. These will be taken up
after two follow-up experiments are reported.
Experiment 2
There was a possible artifactual explanation of the markedly
lower performance in the incidental/base and incidental/im-
posed precise-elaboration conditions relative to the incidental/
base/questioned condition in Experiment
1.
Subjects in the two
nonquestioned conditions may not have processed the sen-
tences for as long as questioned subjects. Incidental subjects in
the nonquestioned conditions were asked to read each sentence
and rate it for comprehensibility, and they did so for every item
without exception within the 7.5-s limit. Subjects in these con-
ditions seemed unhurried. In contrast, subjects in the ques-
tioned conditions were often still trying to come up with an an-
swer at the end of the 7.5-s presentation. A second potential
artifactual explanation of the superiority of the incidental/
base/questioned condition was that the concurrent task of rat-
ing the sentences in the incidental/nonquestioned conditions
may actually have reduced learning
(e.g.,
Postman, 1975).
Because of these possible alternative explanations of the pat-
tern of incidental learning outcomes, we conducted a follow-up
experiment that included the incidental/base/questioned con-
dition as operationalized in Experiment
1
as well as new inci-
dental/base and incidental/imposed precise-elaboration condi-
tions.
In these new conditions subjects were required to process
the sentence "meaningfully" as long as it was presented. They
read each sentence aloud at
a
rate that permitted them
to
under-
stand the sentence, and they were given the emphatic instruc-
tion to make certain that they did in fact understand what they
were reading.
One other question was also addressed in Experiment 2. It
seemed possible that a questioning intervention might produce
gains for the learning of precisely elaborated sentences as well.
If searching for answers to questions about sentences is the rea-
son the incidental/base/questioned condition was so potent in
Experiment
1,
then performance in an incidental/imposed pre-
cise-elaboration/questioned condition might exceed perfor-
mance in the incidental/imposed precise-elaboration condition
and be comparable to the level of recall in the incidental/base/
questioned condition. Thus, subjects in a fourth condition—an
incidental / imposed precise - elaboration / questioned condi-
tion—were asked to tell the experimenter how the precise elab-
oration explained why that particular man did the action speci-
fied for him. We settled on this question after extensive pilot
testing to identify a probe as comparable as possible to the one
used in the base/questioned condition but one that elicited ad-
ditional processing of the sentence. This search for a new ques-
tion was required because the Why did the
particular
man do
that? inquiry used in the base/questioned condition resulted ei-
ther in simple repetition of
the
imposed precise elaboration or
puzzlement when it accompanied sentences containing the im-
posed precise elaborations.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
PRECISE ELABORATIONS295
Table 2
Percentages
of
Questions Answered on Poststudy Test
as a
Function
of
Experimental
Condition:
Experiments 2 and
3
Condition
Base
Base/questioned
Imposed precise-elaboration
Imposed precise-elaboration/
questioned
Experiment 2
(Incidental)
M
18.6
71.0
27.9
56.1
SD
18.1
15.9
14.6
16.1
Experiment 3
(Intentional)
M
54.0
73.0
48.7
49.8
SD
27.4
16.7
17.8
19.1
Note.
Percentages based on 24 responses, n
= 19
per condition.
subjects in either of the nonquestioned conditions, smaller
f(72) = 5.38. Recall levels in the incidental/base and incidental/
imposed precise-elaboration conditions did not differ signifi-
cantly, £(72) = 1.75. In short, the pattern of effects mirrored
Experiment
1
exactly for the three conditions corresponding to
those in the first study. In addition, having subjects answer the
How questions about the precisely elaborated sentences pro-
duced better recall than precise elaborations alone, although
performance in the incidental/imposed precise-elaboration/
questioned condition was inferior to performance in the inci-
dental/base/questioned condition. As in Experiment
1,
supple-
mentary study and posttest measures were revealing about pro-
cessing in these conditions and will be taken up after Experi-
ment
3
is reported.
Method
Subjects. The participants were 76 undergraduates (60 female, 16
male) from the following year's version of the subject pool that provided
subjects for Experiment 1. Participants ranged in age from 17 to 34
years (M
=
20 years 0 months, SD
=
3 years 6 months). Nineteen sub-
jects were assigned randomly to each of the four learning conditions.
Materials. Materials were identical to those used in the correspond-
ing
conditions of Experiment
1.
Subjects in the incidental/imposed pre-
cise-elaboration/questioned condition were presented the precisely
elaborated sentences with the question, "How does the last part of the
sentence make clear why that particular man did what he did?," typed
below the sentence.
Procedures.
The instructions in the incidental/questioned condition
were identical to the instructions in that condition of Experiment 1.
Subjects in the incidental/imposed precise-elaboration and the inciden-
tal/base conditions of Experiment 2 were instructed to read each sen-
tence aloud
as
they heard it, making sure that they understood the sen-
tence.
They
were
told
to keep reading the sentence
out loud for
the
entire
time that it was presented at a rate that permitted them to understand
the sentence. These two conditions were otherwise identical to the cor-
responding condition in the first experiment. Instructions in the inci-
dental/imposed precise -elaboration /questioned condition were the
same as in the incidental/base/questioned condition except that sub-
jects were presented the precisely elaborated versions of the sentences
and were instructed to tell the experimenter how the last part of the
sentence (that
was
not underlined) made clear
why
it
was
that particular
man who did what he did. Testing occurred as it did in Experiment
1,
followed by an interview about strategy use that was similar to the
interview conducted in Experiment 1.
Results and Discussion
As in Experiment 1 the main analysis was on the posttest
recall data, with condition
means
presented in Table
2.
Because
the contrasts of interest could be specified a priori as the six
pair-wise comparisons between the four condition means, these
six planned comparisons were made using the Dunn-Bonfer-
roni approach to error allocation. The overall Type I rate of
.05
was adopted which is comparable to the omnibus overall Type
I rate in the one-way ANOVA alternative analysis. With this
setup,
the per comparison Type I rate was
<.
.0083,
with the
cutoff t(72) - 2.70. Incidental/base/questioned subjects re-
called more than subjects in each of the other conditions, small-
est pair-wise t(12) = 2.83 (MSe - 15.12). Incidental/imposed
precise-elaboration/questioned subjects recalled more than
Experiment 3
That answering questions about precisely elaborated sen-
tences promoted learning under incidental conditions in Exper-
iment 2 raised the possibility that it would also do so when
learning
was
intentional. Thus, the three intentional conditions
of Experiment
1
were replicated and contrasted with an inten-
tional/imposed precise-elaboration/questioned treatment.
Method
Subjects. A total of 76 undergraduates (53 female, 23 male) were
drawn from the same subject pool that provided participants in Experi-
ment 2. Participants ranged in age from 17 to 35 years (M = 19 years
6 months, SD = 2 years 10 months). Nineteen subjects were assigned
randomly to each of the four treatment conditions.
Materials and
procedures.
Materials were identical to those used in
Experiment
2.
Procedures in the intentional/base, intentional/imposed
precise-elaboration, and intentional/base/questioned conditions were
the same as those used in the comparable conditions of Experiment 1.
The instructions in the intentional/imposed precise-elaboration/ques-
tioned condition were the same
as
those used in the incidental/imposed
precise-elaboration/questioned condition of Experiment
2,
with
one
ex-
ception: Subjects in Experiment 3 were told before the sentences were
presented for study about the memory test that would follow presenta-
tion of
the
sentences, and it was assured that they understood that the
goal of their activity
was
to prepare for this memory test
(i.e.,
they were
given intentional learning instructions).
Results and Discussion
The posttest recall data
are
recorded in Table
2.
As in Experi-
ment
2,
all
six
pair-wise comparisons
were made
between means
using the Dunn-Bonferroni procedure with overall Type I
rate
<,
.05 and the per comparison rate
<,
.0083,
with the cutoff
/(72) = 2.70. The intentional/base/questioned mean
was
higher
than each of the other means, smallest t(12) = 2.84. None of
the other three pair-wise contrasts were significant, largest
/(72) = 0.78.
In
brief,
there was exact replication of the pattern of mean
differences between intentional conditions corresponding to
ones in Experiment 1. Although there was an advantage when
questioning was added to imposed precise elaborations under
incidental learning instructions (Experiment
2),
no such advan-
tage
was
detected for intentional learning.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
296PRESSLEY, McDANIEL, TURNURE, WOOD, AHMAD
Supplementary Analyses and Measures:
Experiments 1-3
Supplementary
Across-Experiment
Analyses
of Recall Data
The designs and analyses of the experiments reported here
were sensitive only to large effects
(i.e.,
more than one standard
deviation). As one reviewer pointed out, the power for deter-
mining pair-wise differences could be improved by conducting
analyses by collapsed across experiments. Such supplementary
analyses do in fact strengthen the case that the positive effect of
imposed precise elaborations occurs
when
learning
is
incidental
but not when adults are intentionally trying
to
acquire informa-
tion. When the incidental/imposed precise elaboration versus
incidental/base difference is tested by combining the data of
Experiments
1
and 2 (i.e., by constructing an error term based
on the two relevant cells of Experiment
1
and the two relevant
cells of Experiment 2), the contrast is significant, t(16) = 2.80,
p < .01 (MSe = 210.48). In contrast, when an analogous test
was conducted on the intentional/imposed precise elaboration
versus intentional/base contrast, the trend favoring the base
condition
was
still not statistically reliable,
t(16)
= 1.76, p
>
.05
(MSC
=
475.07).
Supplementary Measures
In each of the three experiments supplementary data were
collected during study in some conditions (answers to the ques-
tions posed during study) and after testing in other conditions
(self-reports about elaboration). As Stein and Bransford (1979)
have noted, elaborative answers to questions and self-reported
elaborations may only scratch the surface of the complete elab-
orative processing. Given this constraint, caution is warranted
in interpreting
these
measures,
although
we
believe,
as do
others
(e.g., Butterfield, Siladi, & Belmont, 1980; Ericsson & Simon,
1983),
that these types of measures often provide more direct
evidence of processing that do performance data alone. Thus,
we include here the most revealing of
these
analyses, but em-
phasize the supplementary nature of these data.
Before turning to the actual results, a few points about these
analyses are in order. Two raters agreed on more than 90% of
the classifications in each of these supplementary analyses, and
disagreements were resolved by discussion. Because between-
conditions patterns were roughly comparable in corresponding
conditions of
the
three experiments, and so that the summary
could be succinct, the supplementary data that are reported in
Tables 3-5 and in the running text were collapsed across com-
parable conditions of the three
studies.
Whenever within-exper-
iments inferential statistical tests were possible, however, they
were conducted and reported. Only effects that were significant
at the p < .01 level were interpreted as significant in these
analyses.
Answers to Why questions during study in the base/ques-
tioned
conditions
(Experiments 1, 2, & 3). The answers pro-
vided by subjects in the base/questioned conditions were ana-
lyzed to determine the likelihood that precise elaborations
would be provided as responses to the Why questions, with the
percentages of
precise
elaborations,
imprecise
elaborations,
and
Table 3
Types
of
Answers Provided During
Study and Associated
Conditional Probabilities
of Recall
in
the
Base/Questioned Conditions
Condition
Incidental/base/
questioned8Intentional/base/
questioned1*
Type of
answer%of
answers
Conditional
recall
probability%of
answers
Conditional
recall
probability
Precise
elaboration
Other
elaboration
No
response
48.2
27.7
24.1
.734
.683
.602
49.3
28.5
22.2
.807
.723
.659
a Based on 39 subjects collapsed over Experiments
1
and 2. b Based on
39 subjects collapsed over Experiments
1
and 3.
no responses recorded in Table 3. The distributions of answer
types were very similar in the intentional and incidental condi-
tions.
Formal inferential analyses were possible between the in-
tentional and incidental conditions of Experiment 1. The per-
centages of the three types of answers did not vary between
these conditions, all pair-wise /s(38) < 1.00.
The conditional probabilities of recall on the memory test
were calculated for each of these three mediators, and these are
also reported in Table
3.2
The conditional probabilities ordered
the same in the intentional and incidental conditions. The high-
est conditional probabilities were observed when the generated
elaborations were precise, although differences in probabilities
of recall were not very great as a function of type of answer
provided during initial item presentation.
Answers to Why
questions during
study in the
imposed-pre-
cise
elaboration/questioned conditions
(Experiments 2 & 3).
Answers were analyzed to determine if responses to the ques-
tions posed in these conditions indicated processing of the pre-
cise elaboration and the type of man. That is, did questioning
produce the processing it was intended to produce? Items were
scored as processed if answers included reiteration of the pre-
cise elaboration and the type of man (e.g., for Sentence 9 in the
Appendix,
being smart
he made
a wise
investment);
reiteration
of the precise elaboration, the type of man, and an appropriate
extension; or an appropriate extension that follows from the
type of man and the precise elaboration (e.g., for Sentence 19
in the Appendix,
getting
tired of
long
hair),
with the percentage
2 Formal inferential statistics were not possible with the conditional
probability data. This was because there were many "missing data"
points
(i.e.,
some subjects offered no responses in some categories) that
analyses often could be conducted only by ignoring important test as-
sumptions, and in most cases this would have been a gross violation.
The situation was improved little by reanalyzing these data with items
as
the random effect, for there were items that failed to produce data in
particular
categories.
Thus, we present only the descriptive conditional
probability data throughout these supplementary analyses.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
PRECISE ELABORATIONS297
Table 4
Percentages
of
Precise Elaborations
Processed and
Not
Processed
and
Associated Conditional Probabilities
of Recall
in
the
Imposed
Precise-Elaboration/Questioned Conditions
Incidental/ Intentional/
imposed precise- imposed precise-
elaboration/questioned" elaboration /questioned*
Processing
status%of
sentences
Conditional
probability
of recall
Conditional
% of probability
sentences of recall
Processed
Not processed81.1
18.9.538
.66778.1
21.9.565
.580
1
n
=
19. Experiment 2.b n
=
19,
Experiment 3.
of processed and nonprocessed items recorded in Table 4 along
with the conditional probabilities of recall on the
test.
The most
important conclusion that follows from these data
was
that sub-
jects usually responded to the questions as they were instructed
to
do
so.
Posttest
interview
data in the base
conditions
(Experiments
I, 2, and
3).
After taking the cued-recall test, subjects in the
base conditions were presented with six of the sentences they
had studied and
were
ask to report what they had been thinking
as the item
was
presented. Each self-report was scored as a pre-
cise elaboration, an imprecise elaboration, or no response. The
percentage of self-reports that fell into each of these categories
is summarized in Table
5.
The formal statistical tests in Experi-
ment I confirmed that there were significantly more precise
elaborations reported in the intentional/base than in the inci-
dental/base condition of that experiment, ;(38) =
3.62,
p
<
.001
(MSC = 162.77), The production of imprecise elaborations did
not differ at the
.01
level between
these two
conditions in Exper-
iment
l,r(38)
= 2.20,
.01
<p<.
03.
The conditional probability
of recall of the posttest probed items as a function of type of
mediator reported is also recorded in Table 5. The conditional
probability differences were not very great for items that were
reported
as
precisely elaborated versus those that
were
reported
as elaborated in some other fashion.
Following the posttest probe items, precise elaboration was
described to subjects. Participants were then asked to estimate
how
many of the
24 items
had been precisely elaborated. Across
experiments, the mean estimate was
14.65%
(3.52 items) in the
incidental/base conditions; and 32.47% (7.79 items) in the in-
tentional/base conditions. The within-Experiment 1 contrast
was significant, f(38) =
3.38,
p < .001
(MS*
=
510.22).
Posttest interview
data in the
(nonquestioned)
imposed-pre-
cise elaboration
conditions.
A
number of analyses of these data
were performed to determine if there was more processing of
precise elaborations
(e.g.,
noting a relation between the subject
of the sentence and the precise elaboration beyond simple repe-
tition of the sentence) when learning
was
intentional compared
to when it
was
incidental.
No means are reported for
these
anal-
yses because level of performance varied tremendously depend-
ing on the criterion adopted for what constituted more than a
repetition. Regardless of the criterion, however, there were only
nonsignificant trends in the direction of greater processing in
the intentional conditions.
Summary. The supplementary measures generally comple-
mented the learning performance data that were the concern in
the main analyses. The incidental versus intentional learning
performance difference was not very great in the base/ques-
tioned conditions; the supplementary data were consistent,
with Why questions (during study) answered in a similar fash-
ion in the incidental and intentional versions of
the
base/ques-
tioned conditions. Just as there was little performance differ-
ence between the incidental and intentional versions of the im-
posed precise-elaboration/questioned conditions, there was
little difference detected in the quality of answers provided in
those conditions. On the other hand, the large learning perfor-
mance difference between incidental/base and intentional/base
conditions was paralleled by a large difference in self-reported
processing between these conditions. The somewhat smaller in-
tentional over incidental performance difference in the non-
questioned imposed precise-elaboration conditions was paral-
leled by reports of slightly more processing in the intentional
compared to the incidental condition. In short, there were clear
parallels between the learning performance results and the sup-
plementary measures believed to provide more direct measures
of process.
A consistent outcome in these supplementary analyses was
that the conditional probabilities of recall for materials that
were precisely elaborated by subjects did not differ much from
the conditional probabilities of recall for materials that subjects
elaborated imprecisely, although the trends usually favored re-
call of precisely elaborated
content.
These
conditional probabil-
ity data, when combined with the modest gains produced by
imposed precise elaborations suggest that possession of a pre-
cise elaboration is not a highly potent variable compared to
other factors that can affect learning in this paradigm
(e.g.,
just
trying to generate a precise elaboration, whether one comes up
with one or not).
General Discussion
Imposed precise elaboration facilitated memory of arbitrary
relations relative to the base sentence condition when learning
Table 5
Types
of
Mediators Reported
and
Associated Conditional
Probabilities
of
Recall in
Base
Conditions
Type of
mediator
Precise
elaboration
Other
elaboration
No mediator
reported
Incidental"
%
reported
4.2
34.2
61.5
Conditional
probability
of recall
.300
.300
.020
Intentional**
%
reported
18.8
40.6
40.6
Conditional
probability
of recall
.773
.695
.284
a n =
39
collapsed across Experiments
1
and
2.
b n
=
39 collapsed across
Experiments
1
and 3.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
298PRESSLEY, McDANIEL, TURNURE, WOOD, AHMAD
was incidental
but
not when
it
was intentional. The facilitation
produced by imposed precise elaborations, however, was mod-
est relative
to
that caused
by
other factors. Most striking,
the
evidence
was
overwhelming that answering
Why
questions
about base sentences produced greater learning than reading
imposed precise
elaborations.
Even the addition of question an-
swering
to
imposed precise elaborations
was not
sufficient
to
bring performance
to the
level produced
by
question-induced
generation
of
elaborations
for
base sentences. These general
outcomes will be elaborated by focusing on the two main issues
addressed by the experimental manipulations studied here.
The Effects of Imposed Precise Elaborations
Consistent with previous reports (e.g., Stein
&
Bransford,
1979;
Stein
et
al., 1984),
the
memory differences between
im-
posed precise elaboration
and
base conditions were
in
favor of
the imposed precise-elaboration condition when learning
was
incidental, although these differences were statistically signifi-
cant only by collapsing over Experiments
1
and 2.
The designs
of the individual experiments simply were
not as
sensitive
to
moderate effects (between one half and three quarters of a stan-
dard deviation)
as the
analyses used
in
previous investigations
of precise elaboration (Cohen, 1977).
In
contrast,
the
perfor-
mance
of
the
intentional/base subjects was slightly better than
the learning
of
intentional/imposed precise-elaboration
sub-
jects
in
both Experiments
1
and
3,
although this difference was
not large enough
to
reach statistical significance even
in the
across-experiments analysis.
Examination of the learning performances in light of the
self-
report data provides some explanation
of
the
greater imposed
precise elaboration effect when learning
is
incidental compared
to when
it
is intentional. Much of the imposed precise elabora-
tion over base advantage observed during incidental learning
was wiped
out in the
intentional situation because
of
the
sub-
stantially greater learning performance of intentional/base sub-
jects compared to incidental/base subjects. The higher learning
performance
in the
intentional versus
the
incidental condition
was probably
due to
two factors: (a)
The
intentional/base sub-
jects produced significantly more elaborations than incidental/
base subjects,
as
indicated
by the
differences
in
their self-re-
ported mediators; and (b) the conditional probabilities of recall
were higher
in the
intentional/base than
in the
incidental/base
condition for
all
three categories of mediation recorded in Table
5.
In
short,
the
intentional learning instruction produced
in-
creases
in
both the amount of elaboration and the effectiveness
of processing.
The greater activity
in the
intentional/base compared
to the
incidental/base conditions contrasts with the failure to
find
sig-
nificant differences
in
processing
in the
intentional/imposed
precise-elaboration versus the incidental/imposed precise-elab-
oration conditions.
It
seems likely that
the
smaller intentional
versus incidental learning differences
in the
imposed precise-
elaboration conditions compared
to the
base conditions were
probably
due to
less dramatic processing shifts produced by the
intentional learning instruction in the imposed precise-elabora-
tion compared to the base conditions.
Effects of Answering Questions
The hypothesis that trying
to
generate answers
to
Why ques-
tions would improve recall
of
base sentences
was
supported,
with
the
effect larger among incidental/base conditions than
among intentional/base conditions. There are
a
number of
pos-
sible explanations for
this
outcome.
It
may be
that this generation-of-answers effect
is a
specific
instantiation
of
Slamecka
and
Graf's (1978) generation effect
(see also Jacoby,
1978).
In
the Slamecka and Graf (1978) exper-
iments, participants generated items
in
response
to
stimulus
words: associates of the stimuli, items
in the
same category, op-
posites, synonyms,
or
rhymes. Subjects were presented
the
stimulus
and
generated
the
response given only
the
first letter
(e.g.,
the
pair save-c with
the
requirement
of
generating
a
rhyme). The learning of pairs that required generation
was
con-
trasted with learning
of
pairs that were read
in
their entirety
(e.g.,
save-cave).
In
general, recall
of
generated responses (i.e.,
the second term
of
the paired associate) exceeded recall
of
re-
sponses that were simply read. Recall of the stimulus item that
prompted generation (i.e., the
first
part of the paired associate),
however, was only slightly improved,
if
at all, over recall of the
stimulus item
in the
read condition (Slamecka &
Graf, 1978,
Experiment
5).
In
contrast,
in
the present experiment, memory
for
the
stimulus information (base sentence)
was
enhanced;
memory
for the
generated
information (answers
to why
ques-
tions) was
not
tested
per
se.
Thus,
to the
extent that
our
results
can
be
characterized
as a
generation effect, they suggest that
generation of semantically related elaborations can have robust
mnemonic effects for the stimulus item prompting the elabora-
tion.
One aspect
of
the base/questioned data deserves special
at-
tention. The conditional probabilities of recall
for
nonanswered
items were
not
much lower than those
for
answered items. Just
trying
to
generate an answer
to
Why questions produced greater
than
60%
recall, which
is
higher than
the
overall recall rates
in
any of the nonquestioned base
or
imposed precise-elaboration
conditions.
It
is interesting that this effect
is
also consistent with
an outcome observed
in
other research
on
generation effects.
Slamecka
and
Fevreiski
(1983,
Experiment 1) studied genera-
tion of opposites, given
a
stimulus word
and
some of the letters
of
the
response (e.g., hot-c
,
hot-c
d).
After attempting
generation, subjects
in
this particular experiment were shown
the whole response word (e.g., cold). Again, these generation
procedures were contrasted with simple reading
of
pairs (e.g.,
hot-cold).
Subsequent recall
of
responses
was
independent
of
whether subjects successfully generated
the
answer during
the
original presentation. Both successfully generated responses
and responses that subjects failed to generate were recalled bet-
ter than response words that were only read. After
a
series
of
follow-up experiments, Slamecka
and
Fevreiski (1983)
con-
cluded that generation failures were really incomplete genera-
tions that involved
a lot of
meaningful encoding
and
elabora-
tion of the stimulus words. Thus,
a
reasonable explanatory hy-
pothesis
for
the present facilitation produced by questioning
is
that subjects were always searching long-term memory
for an-
swers
and
usually made progress toward that goal. Sometimes
their progress was
not
complete enough
to be
formulated into
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
PRECISE ELABORATIONS299
an overt answer to the Why question, but was enough to in-
crease memory of the man-action association.
The most obvious alternative to the generation-effect expla-
nation
is
that generated elaborations were more congruent with
the knowledge base of the learners than were experimenter-cre-
ated, externally imposed elaborations. Imposed precise elabo-
rations can be effective only if they activate knowledge in the
learner that makes the to-be-learned arbitrary relations more
meaningful. Because imposed precise elaborations come from
an external agent's knowledge base, their effectiveness depends
largely on congruence between the prior knowledge of
the
per-
son providing the elaboration and the
learner.
Such congruence
is
hardly guaranteed given the diversity of backgrounds of adult
learners. One part of
the
present results seems consistent with
this interpretation; The imposed precise-elaboration/ques-
tioned subjects performed at a lower level than the base/ques-
tioned subjects. Whatever knowledge the learners possessed
about the imposed precise elaborations should have been acti-
vated and reflected on in the imposed precise-elaboration/ques-
tioned conditions. That activation of this knowledge did not
produce learning as good as that resulting from generation of
the learners' own
precise
elaborations may be because the learn-
ers'
knowledge of imposed precise elaborations was not as rich
as the knowledge of relations produced by the learners them-
selves.
There exist, however, a number of other possibilities that may
account for at least part of the facilitation that was observed in
the base/questioned condition; (a) Attempting to generate an
answer may increase the involvement of consciousness, height-
ened arousal, and/or cognitive effort, which would increase
memory (cf., Jacoby, 1978; Tyler, Hertel, McCallum, & Ellis,
1979);
(b) attempting to generate an answer may force a more
distinctive or deeper encoding of the base relation, thus enhanc-
ing subsequent retrieval (cf. Slamecka
&
Graf,
1978);
(c) gener-
ation itself entails more profound processing than reading (cf.
Slamecka &
Graf,
1978); (d) answering a Why question in-
volves processing that is better matched to the particular crit-
erial task used in these experiments, answering test questions
(cf. McDaniel, Friedman,
&
Bourne,
1978;
Morris, Bransford,
& Franks, 1977); or (e) attempting to answer a Why question
may activate more links in the associative network, which
would provide more routes at retrieval (cf. Anderson & Reder,
1979).
As the current results do not provide unequivocal sup-
port for any one of these hypotheses, more lengthy exposition of
the differing positions would be premature. Further theoretical
development must await studies explicitly designed to test ex-
planations of the current question-answering effects.
In closing, we emphasize that the experiments reported here
illuminate an important production deficiency or inefficiency
(Flavell,
1970)
present in adult
learners.
When facing a difficult
prose task that is
filled
with mutually interfering items, adults
do not spontaneously produce elaborations that are as effective
as ones that they can be cued to generate in response to Why
questions. This processing shortcoming is consistent with elab-
orative production deficiencies and inefficiencies evidenced by
adults in more basic associative tasks in which adults often fail
to produce effective mediators if they try to produce them at
all (e.g., Beuhring & Kee, in press; Pressley & Ahmad, 1986;
Rohwer, Raines,
Eoff,
&
Wagner,
1977).
References
Anderson, J. R., & Reder, L. M. (1979). An elaborative processing ex-
planation of depth of
processing.
In L. W. Cermak &
F.
I. M. Craik
(Eds.),
Levels
of
processing
in human memory
(pp.
385-404). Hills-
dale,
NJ: Erlbaum.
Beuhring,
T.,
&
Kee,
D.
W.
(in
press).
Elaboration and associative mem-
ory development: The metamemory link. In M. A. McDaniel & M.
Pressley (Eds.), Imagery and
related mnemonic
processes:
Theories
and
applications.
New
York:
Springer-Verlag.
Bransford, J. D., & Stein, B. S. (1984). The
ideal problem
solver.
New
York:
Freeman.
Bransford, J. D., Stein, B. S., Vye, N. J., Franks, J. J., Auble, P. M.,
Mezynski, K. J.,
&
Perfetto, G. A. (1982). Differences in approaches
to learning: An overview.
Journal
of
Experimental
Psychology:
Gen-
eral
111,
390-398.
Buium, N., & Tumure, J. E. (1977). A cross-cultural study of verbal
elaboration productivity and memory in young children.
Child
De-
velopment,
48, 296-300.
Butterneld, E. C, Siladi, D., & Belmont, J. M. (1980). Validating theo-
ries of intelligence. In H. W. Reese & L. P. Lipsett (Eds.),
Advances
in
child development
and
behavior
(Vol.
15, pp. 95-162). New York:
Academic Press.
Cohen, J. (1977). Statistical
power
analysis for the
behavioral
sciences.
New
York:
Academic Press.
Ericsson, A., & Simon, H. A. (1983).
Verbal protocol
analysis.
Cam-
bridge, MA: MIT Press.
Flavell, J. H. (1970). Developmental studies of mediated memory. In
H. W. Reese & L. P. Lipsett (Eds.),
Advances
in child
development
and
behavior (Vol.
5,
pp. 182-211). New
York:
Academic Press.
Franks, J. J.,
Vye,
N. J., Auble, P. M., Mezynski, K. J., Perfetto, G. A.,
Bransford, J. D., Stein, B. S., & Littlefield, J. (1982). Learning from
explicit versus implicit texts. Journal of
Experimental
Psychology:
General,
111,406-422.
Glover, J. A., Bruning, R. H., & Plake, B. S. (1982). Distinctness of
encoding and
recall
of text
materials.
Journal
of
Educational
Psychol-
ogy,
74,
522-534.
Graf,
P. (1980). Two consequences of
generating:
Increased inter- and
intraword organization of sentences.
Journal
of
Verbal Learning and
Verbal
Behavior,
19,
316-327.
Graf,
P.
(1982).
The memorial consequences of generation and transfor-
mation. Journal of
Verbal Learning
and
Verbal
Behavior,
21, 539-
548.
Hashtroudi, S., Parker, E. S., DeLisi, L. E., & Wyatt, R. J. (1983). On
elaboration and alcohol.
Journal
of
Verbal Learning
and
Verbal
Be-
havioral,
164-173.
Hedges,
L.
V.,
&
Olkin,
I.
{1985).
Statistical
methods for
meta-analysis.
Orlando, FL: Academic Press.
Jacoby, L. L. (1978). On interpreting the effects of
repetition:
Solving
a
problem versus remembering a solution.
Journal
of
Verbal Learning
and
Verbal
Behavior,
17,
649-667.
Kirk, R. E. (1982). Experimental design (2nd ed.). Belmont, CA:
Brooks/Cole.
Levin, J. R.
(1976).
What
have we
learned about maximizing what chil-
dren can learn? In J. R. Levin
&
V.
L.
Allen
(Eds.),
Cognitive learning
in
children
(pp.
105-134). New
York:
Academic Press.
McDaniel, M. A. (1984). The role of elaborative and schema processes
in story memory.
Memory
and
Cognition,
12,
46-51.
McDaniel, M. A., Friedman, A., & Bourne, L. (1978). Remembering
the
levels
of information in
words.
Memory &
Cognition,
6,
156-164.
McDaniel, M. A.,
&
Kearney,
E.
M.
{1984).
Optimal learning strategies
and their spontaneous
use:
The importance of task-appropriate pro-
cessing.
Memory &
Cognition,
12,
361-373.
Morris,
C.
D., Bransford,
J.
D.,
&
Franks,
J. J.
(1977).
Levels of process-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.