Você está na página 1de 5

Journal of

Experimental Psychology
VOL. 79, No. 1, PART 1 JANUARY 1969

INSTRUCTIONS TO USE VERBAL MEDIATORS IN


PAIRED-ASSOCIATE LEARNING 1
MARIAN SCHWARTZ
University of Wisconsin

An instructional set to use verbal mediators was compared with standard


instructions in its effect on the acquisition of a single paired-associate (PA)
list. Two experiments were run in which the study trial was paced at a
2-sec. rate. Recall was paced at 2 sec. in the first experiment and was
untuned in the second. Instructions to mediate greatly increased the number
of reported mediators in both experiments, but facilitated performance only in
the second. The results of Exp. II were interpreted as evidence that
natural language mediators can play a causal role in facilitating PA
learning by college level .S's. The lack of facilitation in Exp. I was tentatively
attributed to inability to utilize mediators effectively with 2-sec. recall
intervals. It was suggested that the materials and instructions employed may
allow greater experimental control than a transfer design typically affords
over the mediators which Ss actually use.

The use of what has variously been called In order to impose some systematic control
mnemonic devices, associative aids, natural on S's use of mediators in the acquisition of
language mediators, or implicit mediators has a single paired-associate (PA) list, several
been found to be related to better perform- investigators have employed instructional
ance in the acquisition and retention of paired sets. Jensen and Rohwer (1963) instructed
associates (e.g., Montague, Adams, & Kiess, adult retardates to make up a sentence or
1966; Runquist & Farley, 1964). That is, phrase linking the two members of each pair,
pairs which, according to S's report, were and found that those instructed to mediate
learned with the aid of a mediator tend to in this way learned far faster than did un-
be learned faster and retained better than instructed controls. Retardates were used
pairs for which 5" reports no mediator. That by these investigators on the assumption that
the use of mediators facilitates acquisition and such 5"s would be likely to form verbal
retention does not necessarily follow, how- mediators in rote learning if, and only if,
ever, from this observed relationship. The instructed to do so by E; whereas normal
reported mediator may be a correlate or a adults, particularly college students, would
consequence, rather than a determinant, of be expected to mediate whether instructed
high associative strength. to do so or not.
. , . Several recent experiments have compared
^s^^K^dS^^ the effects of instructions to mediate vs.
Grant to the Graduate School of the University of standard instructions, using Ss at the college
Wisconsin. or high school level. With such .S's, clear
1
1969 by the American Psychological Association, Inc.
MARIAN SCHWARTZ

evidence that mediation instructions facili- 1967). It was also assumed that the phras-
tated learning was found only when SB were ing of the instructions and the nature of the
told to use mediating images, not when they materials used would be important factors in
were instructed to use verbal mediators or determining whether learning would be facil-
when the nature of the mediator to be em- itated as a result of mediation instructions;
ployed (imaginal or verbal) was left unspeci- accordingly, materials and instructions were
fied. In a study by Martin and Dean designed to maximize the possibility of facili-
(1966), instructions to "associate or link up tation.
the nonsense syllable and the word in some
way" did not significantly affect PA learning. EXPERIMENT I
In three experiments by McNulty (1966), Method
instructions to form ridiculous or bizarre
Materials.A list of pairs was constructed on
linkages between the stimulus and response the assumption that instructions to use a verbal
terms had no significant effect, although the mediator would be most effective, relative to a
direction of the learning differences consist- control receiving standard instructions, where (a)
ently favored the mediation as opposed to associations between stimulus and response terms
the standard condition. Bugelski, Kidd, would be relatively hard to form directly, without
the use of mediators, (6) a potentially useful medi-
and Segmen (1968) facilitated recall of ator exists for each pair, but (c) the potential
object-name responses by teaching 5s to as- mediators are not so obvious that they would be
sociate these responses with the Stimulus as readily available to uninstructed as to instructed
Numerals 1-10, using the technique of "pic- ,?s. The materials consisted of the following 10
pairs, in which the stimulus was a single letter and
turing" the responses together with mediators the response was a word: A-pie, j>-ca/t, E-hard, o-
which rhymed with the numerals. Yarmey bad, K-murder, i*-hate, 'B.-jast, t-clock, v-win,
and Thomas (1966) found that instructions X-WMJIC. Each response term is a primary re-
to mediate via mental images facilitated sponse to a word which is itself a primary response
learning of concrete nouns, but instructions to to the stimulus letter. According to the norms pub-
lished by Anderson (1965) for single associations
use verbal mediators (to form "silent verbal to uppercase letters, the primary responses to the
chains" connecting the members of each pair) 10 stimulus letters used are apple, dog, easy, good,
inhibited learning of abstract noun pairs. In kill, love, run, time, victory, and xylophone, re-
a study by Paivio and Yuille (1967), 5s spectively. To obtain the response terms for the
present study, the 10 words apple, dog, etc., were
were instructed to use either verbal medi- included among the stimuli in a discrete word
ators, mental images, or rote repetition in association test administered to an undergraduate
learning a PA list. Recall was poorer for psychology class; and the primary responses to
the rote-repetition group than for either of these 10 words, based on data from 38 undergradu-
the two mediation groups, a result which ates, were used as the responses in the PA list
shown above.
could have been due to facilitation in the Procedure.The pairs were presented on a
mediation conditions, inhibition produced Lafayette memory drum for a single study trial in
by the repetition instructions, or both. which each pair was shown for 2 sec., followed by
The present study investigated the effect of a single recall trial in which each stimulus letter
was presented alone for 2 sec. The 6"s were run
an instructional set to use verbal mediators individually. They were instructed to study the
on the acquisition of a single PA list by col- pairs on the study trial, and to call out the response
lege undergraduates. The main purpose was words when the single letters appeared alone on
to manipulate as an independent variable the the recall trial. In addition, 5s in Group I (in-
use of natural language verbal mediators in structed), but not those in Group U (uninstructed),
were told that "some subjects find it easier to learn
order to determine whether such mediators a list of this type by mentally changing the left-
can play a causal role in facilitation of PA hand letter into a word that begins with that letter
learning by college level 5s. It was ex- and is also related to the right-hand word," and
pected that substantial mediation would occur were advised that they might find it easier to use
even in the absence of an instructional set, the same strategy. Between study trial and recall
trial there was an 8-sec. interval during which S
but that the number and type of mediators was reminded of the recall instructions. One
employed by 5s could be partially controlled minute after the end of the recall trial, 5" was given
through instructions (cf. Paivio & Yuille, a sheet on which were the 10 stimuli along with his
INSTRUCTIONS TO USE VERBAL MEDIATORS

own response, if any, to each. For all pairs, re- (50) = .86, p > .10. This combination of
gardless of whether a response had been given, resultsthat instructions to mediate increase
5 was asked to write whatever associations or aids
to memory he had used to help him remember, or the number of pairs on which 5"s report using
try to remember, the responses. mediators, that learning is superior for re-
Serial positions and sequences of items were portedly mediated pairs, but that instruc-
changed between study trial and recall trial, and at tions to mediate have no significant effect on
least four other items intervened between the study group differences in learningis completely
trial and recall trial appearance of any given
item. For generality, two different sets of study consistent with the findings of Martin and
and recall orders were used. Dean (1966).
Subjects.The SB were 52 undergraduates from
an elementary psychology course, none of whom EXPERIMENT II
had taken the word association test described above.
The 5s were assigned to conditions in blocks of Method
2, with 1 instructed and 1 uninstructed S per block.
Running order within each block was random. The failure of the instructional set to produce
a difference between groups in recall perform-
ance may have been due to the form or the rapid
Results pacing of the recall test, or to the wording of the
instructions. With the idea of maximizing the
A count was made of the number of pairs chances of producing facilitation with mediation
on which 5 reported using, or tried to use, a instructions, the experiment was rerun with the
stimulus letter (SL) mediator; i.e., a medi- following changes; (a) The recall test was untimed.
ator which consisted of a word starting with Immediately after the study trial, S was given a
sheet on which were the 10 stimulus letters, and
the stimulus letter for that pair. The cate- told to write after each letter the word that had
gory of "SL mediator" corresponded to the been shown with it. (6) The recall test was
type of mediator which the Group I in- followed by a matching test, also untimed, in which
structions encouraged 5"s to use; it included S was shown the 10 stimulus letters and an alpha-
but was not limited to the primary responses betical list of the 10 response words, and required
to write each word after the appropriate stimulus.
to the stimulus letters as shown in the Ander- Immediately after the matching test, S was
son (1965) norms. These primary re- asked to report whatever mediators he had used in
sponses were found to constitute 85.8$; learning, (c) Examples were added to the media-
of all SL mediators and were not separately tion instructions for Group I. Two sample pairs
(N-silent and o-salt) were shown, and Group I
analyzed. 5s were told that they might mentally change the N
Stimulus letter mediators were reported to noisy, the o to ocean. These pairs were also
for a mean of 4.69 pairs in Group I and 2.35 incorporated in the instructions for Group U as
pairs in Group U. The difference was sig- examples of the type of pair 5s would be asked
to learn, but with no mention of possible mediators,
nificant, * (50) = 3.16, p < .01. A relation- (rf) The instructions before the study trial in-
ship between reported mediation and correct cluded the warning that each pair would be shown
recall is indicated by the finding that 74.3% for only a couple of seconds.
of all pairs with and 21.7% of all pairs with- Subjects.The 5s were 32 undergraduates from
out SL mediators were correctly recalled. If an elementary psychology course, none of whom
had served in Exp. I or in the word association
"mediator" is defined so as to include not test. Assignment of 5s to groups was as in Exp. I
only SL mediators but any kind of associa- with the additional restriction that 5s with previous
tive aid reported by S, the mean number of experience in verbal learning experiments should
mediated pairs would be 5.31 in Group I be equally represented in the two groups.
and 3.54 in Group U, * (50) = 2.59, p < .01;
and the percentage of pairs correctly recalled Results
would be 74.8% of all mediated and 12.8% The mean number of correct responses on
of all nonmediated pairs. the recall test was 8.56 in Group I and 6.00 in
A comparison between Group I and Group Group U; on the matching test, mean cor-
U in the number correct on the recall trial rect was 9.56 in Group I and 6.88 in Group
showed no significant difference. The mean U. The number of -S"s out of 16 who gave
number of responses correctly recalled was all 10 responses correctly was, on the recall
4.31 in Group I and 3.73 in Group U, t test, 6 in Group I and 1 in Group U; on the
MARIAN SCHWARTZ

matching test, 13 in Group I and 5 in Group The failure of the instructional set to affect
U. Because of the apparent heterogeneity recall in Exp. I and its success in Exp. II
of variance and skewness of the distributions, could have stemmed from several procedural
the Mann-Whitney U test was used to evalu- differences between the two studies. The
most important of these differences, however,
ate significance. Number correct was signifi-
is probably the time allowed for recall. On
cantly greater in Group I than in Group U on the assumption that the increase in recall time
both the recall test, U = 47.5, p < .001, and was the crucial procedural change between
matching test, U = 55.5, p < .01. In view Exp. I and II, the results are consistent with
of the ceiling effect in Group I, particularly those of Schulz and Lovelace (1964), and
on the matching test, these figures undoubt- may be interpreted in terms of their distinction
edly underestimate the difference between between discovery and utilization of mediating
the two groups. associations. Schulz and Lovelace varied the
lengths of the anticipation and study intervals
The absolute number of mediators reported
in the test list of a three-stage mediation para-
was greater than in Exp. I, but the same digm. They failed to find mediation with a
trends emerged in regard to differences in 2 : 2 rate or a 2:4 rate of presentation but
reported mediation between Groups I and U. succeeded with a 4: 2 rate. Their conclusion
The primary responses to the stimulus letters was that lengthening the anticipation interval
accounted for 87.9% of all SL mediators, (i.e., the time allowed for response recall)
and were not analyzed separately. Stimulus provided more opportunity for effective utili-
letter mediators were reported for a mean of zation of already discovered mediators. In
8.75 pairs in Group I and 4.69 pairs in the present study, instructed 5"s in both ex-
Group U, a significant difference, t (30) periments reported that they used, or tried to
use, many more mediators than did the unin-
4.53, p < .001. It was found that 89.8% of
structed 5s. A concomitant difference in
all pairs with and 38.1% of all pairs without learning, however, occurred only in Exp. II,
SL mediators were correctly recalled. If where unlimited time was allowed for recall,
"mediator" is defined so as to include any and not in Exp. I, where the recall test was
kind of associative aid reported by S, the paced at a 2-sec. rate. An interpretation in
mean number of mediated pairs would be 9.00 terms of the discovery-utilization distinction
would be that the instructional set facilitated
in Group I and 6.62 in Group U, t (30) =
discovery of mediators during the study trial,
3.14, p < .005, and the percentage of pairs but that effective utilization of these mediators
correctly recalled would be 85.2% of all was possible only when the time allowed for
mediated and 28.6% of all nonmediated pairs. response recall was longer than 2 sec.
Experiments on mediation in PA learning
DISCUSSION have been of two kinds: (a) transfer studies,
in which one or more stages of the mediation
The major finding of Exp. II is that over- paradigm involve associations built in through
all learning was facilitated by an instructional transfer from prior lists learned in the experi-
set to use verbal mediators. This, coupled mental situation; and (&) single-list acquisi-
with the findings that instructions to mediate tion studies, in which the mediators under in-
increased the number of mediators reported and vestigation consist of natural language verbal
that reportedly mediated pairs were more likely associations or of images, with no experi-
to be learned, provides evidence that natural mentally built-in components.
language mediators can play a causal role in In regard to the transfer approach, several
facilitating PA learning by college level S"s, investigators have recently suggested that hav-
and are not simply correlates or consequences ing 6"s learn a sequence of lists conforming
of high associative strength. An alternative, to an -defined mediation paradigm may afford
although less parsimonious, interpretation less experimental control than had previously
would be that the instructional set facilitated been supposed over the associations which 5s
performance through some nonmediational actually use in learning the test-list pairs
mechanism (e.g., by increasing motivation), (Adams, 1967; Martin & Dean, 1964; Richard-
and that the performance gain effected in this son, 1968), or may not, in fact, produce media-
way produced, in turn, an increase in the tion at all (Handler & Earhard, 1964). In
number of mediators reported. particular, it should be recognized that S1 is
INSTRUCTIONS TO USE VERBAL MEDIATORS

likely to draw upon natural language habits in REFERENCES


forming his own mediators, which he inter- ADAMS, J. A. Human memory. New York:
poses between the stimulus and response terms McGraw-Hill, 1967.
of the pairs he is required to learn, either in ANDERSON, N. S. Word associations to individual
place of or in addition to the -defined media- letters. Journal of Verbal Learning and Verbal
tor (Montague & Kiess, cited in Adams, 1967, Behavior, 1965, 4, S41-S4S.
p. 96). While this tendency to import natural BUGELSKI, R. B., KIDD, E., & SEGMEN, J. Image
language mediators may not bias comparisons as a mediator in one-trial paired-associate learn-
in gross performance between the mediation ing. Journal of Experimental Psychology, 1968,
and control conditions, it may seriously confuse 76, 69-73.
JENSEN, A. R., & ROHWER, W. D., JR. Verbal
any attempts at a fine-grained analysis of the mediation in paired-associate and serial learning.
mediational processes involved in test-list per- Journal of Verbal Learning and Verbal Behavior,
formance when the transfer approach is used. 1963, 1, 346-352.
Single-list acquisition studies have, as a HANDLER, G., & EARHARD, B. Pseudomediation: Is
rule, afforded an opportunity for the use of chaining an artifact? Psychonomic Science, 1964,
natural language or imaginal mediators in 1, 247-248.
learning, and for postexperimental mediation MARTIN, R. B., & DEAN, S. J. Implicit and explicit
reports, without experimental manipulation of mediation in paired-associate learning. Journal
the specific mediators used. of Experimental Psychology, 1964, 68, 21-27.
The present study involves an attempt to MARTIN, R. B., & DEAN, S. J. Reported mediation
in paired-associate learning. Journal of Verbal
exercise experimental control over the use of Learning and Verbal Behavior, 1966, 5, 23-27.
natural language mediators in a single-list ac- McNuLTY, J. A. The effects of "instructions to
quisition experiment. It is, in effect, an A-B, mediate" upon paired-associate learning. Psycho-
B-C, A-C chaining paradigm in which the first nomic Science, 1966, 4, 61-62.
two stages are inferred from word association MONTAGUE, W. E., ADAMS, J. A., & KIESS, H. 0.
norms and the third corresponds to the single Forgetting and natural language mediation.
list presented. This list, particularly in con- Journal of Experimental Psychology, 1966, 72,
junction with the mediation instructions, ap- 829-833.
pears to allow a reasonable degree of experi- PAIVIO, A., & YUILLE, J. C. Mediation instructions
mental control over the class of mediators used and word attributes in paired-associate learning.
Psychonomic Science, 1967, 8, 65-66.
by 5s, and over the specific verbal mediators RICHARDSON, J, Implicit verbal chaining as the
within that class, assuming that the postexperi- basis of transfer in paired-associate learning.
mental reports reflect the associations actually Journal of Experimental Psychology, 1968, 76,
used during learning. In Group I of Exp. II, 109-115.
e.g., 97% of all associative aids reported were RUNQUIST, W. N., & FARLEY, F. H. The use of
SL mediators, and 89% of these SL mediators mediators in the learning of verbal paired asso-
were the primary responses to the stimulus ciates. Journal of Verbal Learning and Verbal
letters as shown in the Anderson (1965) norms. Behavior, 1964, 3, 280-285.
While the technique of the present study does SCHULZ, R. W., & LOVELACE, E. A. Mediation in
not provide complete experimental control over verbal paired-associate learning: The role of
the mediators used, it probably provides a temporal factors. Psychonomic Science, 1964,
1, 95-96.
greater degree of control than occurs when the
YARMEY, A. D., & THOMAS, K. A. Set and word
mediating association which E attempts to abstractness-concreteness shift in paired-associate
build in through training must compete with a learning. Psychonomic Science, 1966, 5, 387-388.
natural language mediator imported into the
experimental situation by 5". (Received February 7, 1968)

Você também pode gostar