You are on page 1of 12

Grasping the Meaning of Metaphor: Story Recall and Comprehension

Author(s): John E. Waggoner, Miriam J. Messe and David S. Palermo


Source: Child Development, Vol. 56, No. 5 (Oct., 1985), pp. 1156-1166
Published by: Wiley on behalf of the Society for Research in Child Development
Stable URL: http://www.jstor.org/stable/1130230
Accessed: 30-09-2017 03:46 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/1130230?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Society for Research in Child Development, Wiley are collaborating with JSTOR to
digitize, preserve and extend access to Child Development

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
Grasping the Meaning of Metaphor: Story Recall
and Comprehension

John E. Waggoner, Miriam J. Messe, and David S.


Palermo

Pennsylvania State University

WAGGONER, JOHN E.; MESSE, MIRIAM J.; and PALERMO, DAVID S. Grasping the Mea
Metaphor: Story Recall and Comprehension. CHILD DEVELOPMENT, 1985, 56, 1156-1166. Ch
ability to recall and explain the meanings of metaphors was investigated. Metaphors were em
in stories at the Reaction or Outcome positions using a story grammar structural form. R
metaphors was equal to recall of literal statements with comparable meanings by 7-, 9-, and
old children. Recall was better if the statements were in the Outcome node of the story than it
they were in the Reaction node. Probes for the meaning of the metaphors, however, revea
comprehension of metaphors in both positions. Metaphors had no differential effect on
subsequent parts of the story. The results are discussed in terms of the relations between
and the construction of meaning for both literal and figurative language.

late concrete operational stage fail to com-


Bransford and McCarrell (1974) proposed
prehend metaphors (Asch & Nerlove, 1960;
that the process of comprehending metaphor
is identical in kind to the process ofBillow,
com- 1975; Cometa & Eson, 1978). Vos-
prehending literal language. In both niadou
cases, et al. (1984) suggest that these findings
the comprehender must specify a context can be
or attributed to methodological shortcom-
ings that introduce too many irrelevant
situation within which the relations specified
by the linguistic input can make sense. sources
Al- of difficulty into the experimental
though Bransford and McCarrell did not situation.
dis- For instance, many studies have
cuss how or whether the ability to presented
com- children with metaphors in isola-
prehend metaphor changes over the coursetion, of
a procedure that not only is highly atyp-
development, their position implies thatical ofanormal language use but also ignores
the findings that adults sometimes have
child who possesses the ability to understand
literal language should be able to demon-
difficulty comprehending literal language
strate understanding of metaphor aswhen
well.contextual support is lacking (e.g.,
This position has recently been taken by
Bransford & Johnson, 1972). The validity of
Palermo (1982, in press) and Vosniadou,
thisOr-
criticism is demonstrated by recent re-
tony, Reynolds, and Wilson (1984), whosearch
argueindicating that children perform well
that comprehension of both literal and
when metaphors are placed in supportive
contexts (Reynolds & Ortony, 1980; Vos-
metaphoric language involves the derivation
of meaning from linguistic input rather niadou
than a et al., 1984). Vosniadou et al. suggest
simple reading off of the "given" meaningthat context
of aids in comprehension because it
the input. If this is the case, children should
allows the comprehender to make inferences
experience no insurmountable difficulty aboutinthe nature of the material presented.
comprehending metaphor if they can under-Contextual support makes the meaning of the
stand literal language. metaphors predictable. A second source of
difficulty in many studies of metaphor com-
In apparent contrast to the Bransford prehension
and has been that they assess com-
McCarrell position, there has been a substan-
prehension through procedures requiring ver-
tial amount of research that has been taken to bal explication of grounds or paraphrase of
indicate that children below the level of the the metaphor, tasks that tap metalinguistic

The authors wish to express their appreciation to Dr. Ronald W. Gray, superintendent of the
Bellefonte Area School District, Dr. Ott Kantner, supervisor of elementary education, Dr. Betsy
Futterman, principal of the Marion Walker Elementary School, and Mr. Elton Abel, principal of the
Bellefonte Middle School as well as the teachers of grades 2, 4, and 6 in these schools for their
helpful cooperation while this research was being conducted. Requests for reprints should be sent
to David S. Palermo, 441 Bruce V. Moore Building, Department of Psychology, Pennsylvania State
University, University Park, PA 16802.

[Child Development, 1985, 56, 1156-1166. ? 1985 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/85/5605-0005$01.00]

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
Waggoner, Messe, and Palermo 1157
skills as well as basic comprehension abili- sources of difficulty interact, the last sugges-
ties. Research employing nonverbal assess- tion seems the most likely. Verbal assessment
ment techniques (Reynolds & Ortony, 1980; tasks (high difficulty) and nonpredictive con-
Vosniadou et al., 1984; Winner, Engel, & texts (high difficulty) lead to relatively poor
Gardner, 1980) indicates that children as performance, while nonverbal tasks (low
young as 3-4 years of age understand difficulty) and predictive contexts (low
metaphors if appropriate assessments aredifficulty) result in high levels of perfor-
used. In short, children's failure to demon- mance.

strate an ability to comprehend metaphor in


these studies is due to the presence of these The idea that sources of difficulty
and other sources of difficulty. to influence metaphor comprehe
suggests a possible answer to the qu
The suggestion that different sources ofwhy children cannot verbalize the m
difficulty interact to influence metaphor com-metaphors. It may be that the ef
prehension in children provides a fruitfuldifficult task, such as verbalization
framework for viewing much of the work in ing, can be counteracted by decr
metaphor comprehension. It also provides level
a of difficulty of other factors, s
possible answer to several questions. Al- textual support. That is, if predictiv
though the cited studies demonstrate theis employed, a context in which th
facilitative role of context, some studies of the metaphor is determined b
(Gardner, Kircher, Winner, & Perkins, 1975; rounding material, one might be ab
Silverman, Winner, & Gardner, 1976; Win-onstrate high levels of comprehen
ner, Rosenstiel, & Gardner, 1976) have placedthough verbal assessment procedur
metaphors in contexts and reported relativelyployed.
low levels of appropriate responding. Why
context helps in certain situations but not in The fundamental problem to be dealt
with is how to devise predictive contexts for
others needs clarification. Second, why it is so
difficult for children to give a verbal indica-metaphors. In the present study, as in previ-
tion of the meaning of metaphors is also notous research, metaphors were placed in ex-
tended story contexts in order to provide con-
clear. It would seem that, if children are capa-
ble of selecting appropriate interpretations intextual support. Unlike previous research,
however, the stories used were written to
a multiple-choice task (Reynolds & Ortony,
1980; Vosniadou & Ortony, 1983; Winner etconform to the ideal story structure described
by recent research on story schemata (Mand-
al., 1980) or of acting out the meaning of
metaphors in a "toy world" (Vosniadou et al.,ler & Johnson, 1977; Rumelhart, 1975; Stein
1984), they should be able to verbalize, even & Glenn, 1979). The particular story grammar
if only crudely, the meaning of a metaphorused as a basis for writing the stories was that
of Mandler and Johnson (1977). Story
they have been demonstrated to know by
other means. There is no obvious reason whyschemata play important roles in the com-
prehension of stories (Mandler, 1978; Stein,
verbalization should be any more difficult
than, for example, acting out the meaning. 1982). Stories that do not conform to the ideal
story schemata, for example, are more difficult
The first question might be answered byto integrate and understand than those that do
noting that Gardner et al. (1975), Silverman et conform (Mandler, 1982). Thus the use of a
al. (1976), and Winner et al. (1976) did not usepoorly structured story as a context may lead
extended story contexts as support for to less than optimal performance in metaphor
metaphors. Presumably, short sentences do comprehension tasks since understanding the
not provide enough of a base for inferences tostory context requires increased processing
be made about the possible meanings of the capacity, which could otherwise be allocated
metaphors, while extended story contexts do. to metaphor comprehension. One reason,
Thus contextual support aids metaphor com-then, for using stories written to conform to
prehension when it provides a sufficient base story schemata is that failing to follow this
for inference making. There is, however, procedure
a may lead to unwanted decrements
second factor that could account for the differ-in performance. However, a more important
ent results. Vosniadou et al. (1984) used non-reason for the use of stories written in accor-
verbal assessment procedures, while the
dance with story schemata is that these
other researchers employed verbal measures.schemata enable one to make inferences
Poor comprehension could be attributed toabout material appropriate to the story (Mand-
low levels of contextual support, to the use ofler & Johnson, 1977; Whaley, 1981). If, as
verbal assessments, or to both factors interact-
Vosniadou et al. (1984) suggest, the role of
ing. If, as Vosniadou et al. suggest, thecontext is to provide a base for inference mak-

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
1158 Child Development

ing, thereby aiding the comprehension ofMethod


metaphor, then using stories conforming to
story schemata should lead to an increase in Subjects.-Ninety-six children partici-
pated in the study, 16 males and 16 females
metaphor comprehension.
from each of three age levels: second grade,
Two verbal measures of comprehensionmean age 7-11; fourth grade, mean age 9-9;
were taken in the present study. First, as hasand sixth grade, mean age 11-9. The second
typically been done in story comprehension and fourth graders attended an elementary
research (e.g., Mandler & Johnson, 1977; school in a rural community in central Penn-
Stein & Glenn, 1979), the children were sylvania. The sixth graders were students at
asked to recall the stories. In addition to es- the middle school in the same school district.
tablishing whether metaphoric material in the Three fourth and four sixth graders were re-
story would be recalled as well as comparable cruited from another local school district to
literal material, the recall measure allowed di- complete the sample.
rect comparisons with prior research examin-
ing the influence of story grammars on story
Materials.-Eight simple two-episode
stories were constructed in accordance with
comprehension. While story recall has been
demonstrated to reflect the listener's own ex- the story grammar outlined by Mandler and
pression of the gist of the story (Hildyard & Johnson (1977). Each of the eight stories con-
Olson, 1978), this measure has been criticized sisted of two Setting propositions followed by
two causally connected episodes, each con-
as insufficient by itself, at least when a single
story is used (e.g., Omanson, Warren, & taining six nodes with one proposition per
Trabasso, 1978). It has been argued that node. The development of the first episode
rote recall without comprehension can be caused an ending that began the development
achieved for a short story. Thus while rote of the second episode such that the Ending
recall of a series of four stories seemed un- node of the first episode was the same as the
likely in the present study, a second measure Beginning node of the second episode. Each
of metaphor comprehension was employed. story consisted of thirteen propositions: Set-
Following recall of the fourth story in each set
ting (Si), Setting (S2), Beginning (B), Reaction
presented, the children were specifically (R1), Goal (G1), Attempt (A1), Outcome (O1),
Ending/Beginning (E/B), Reaction (R2), Goal
probed for the meanings of each metaphor in
the stories. This procedure was used for sev- (G2), Attempt (A2), Outcome (02), and Ending
eral reasons. First, the metaphors were em- (E). Each of the stories contained a single
bedded in two different structural positions in
metaphorical proposition. In four of the
the grammar of the stories, one known from stories, the metaphor was located at the R1
prior research (e.g., Mandler & Johnson, node. In the other four stories, it was located
in the 01 node.
1977) to be frequently recalled and one
known to be infrequently recalled relative to Each story had a literal version as well as
other propositions in the story. The probe for a metaphoric version. Literal versions were
meaning of the metaphor would, therefore, formed by inserting interpretations of the
help to determine whether material that was metaphors into the stories in the form of lit-
not recalled was meaningful to the child, eral propositions at the target nodes. Thus
even though not recalled. In addition, of though there were only eight story themes,
course, these data would give further informa- there were a total of 16 stories: eight stories
tion on the relation between the two mea- containing a metaphor at one of the target
sures of comprehension. It was also assumed nodes and another eight (the literal coun-
that the children might be able to construct a terparts of the metaphor stories) containing a
meaning for the metaphor from the story con- literal proposition of comparable meaning in
text if they were given a probe following the place of the metaphor at the target nodes. All
story, as demonstrated in previous research. other propositions were held constant. See
Finally, the probe procedure would help in the Appendix for two examples of stories
understanding whether the children under- used with their literal-metaphor counterparts.
stood the meaning of the metaphorical sen-
tences they had recalled in surface or ver- Stories were divided into two sets, A and
batim form. Children would be more likely to B. Each story set contained eight stories: four
comprehend a metaphor if it were recalled in stories with metaphors at the target nodes and
any form than if it were not recalled but even four stories with literal targets. The four meta-
more likely to comprehend a metaphor re- phor stories in Set A (two with metaphors in
called in gist than in verbatim form since re- the R1 position and two with metaphors in the
call in gist form presumably indicates that the O1 position) had their literal counterparts in
metaphor has been comprehended rather Set B. Set B likewise contained four metaphor
than just recalled by rote. stories with their literal counterparts in Set A.

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
Waggoner, Messe, and Palermo 1159

All stories were given a short descriptive titleasked to tell as much as he or she could re-
used to cue recall. member about each story (recall was cued by
story title). After the subject finished recall of
Stories were tape-recorded on individual
a story, he or she was prompted for more re-
cassette tapes (one story, preceded by the ti-
call (e.g., "Did anything else happen?" "Can
tle, per tape) to eliminate variations in voice
you remember any more about the story?").
inflection across subjects. Care was taken to
The order of story recall was the same as the
assure that the propositions had equal empha-
order of presentation. After recall of the
sis across stories and that each story was told
fourth story, the subjects were specifically
in approximately the same length of time.
probed for metaphor comprehension. They
Design.-Factors manipulated included were given the title of the first metaphor story
type of target (metaphor vs. literal target prop-they heard, were asked if they remembered
osition), position of target in the story (R1 vs.the target, and were asked for its meaning
01 node), and grade level (second, fourth, and(e.g., "In the story [title], do you remember it
sixth grades). An additional factor was createdsaid [metaphor]? What does it mean?"). The
by grouping the stories into two sets (A and same procedure was followed for the second
B). Type of target and position of target metaphor target. After the subjects' first re-
served as within-subject factors. sponse they were asked if they could think of
any other meaning of the sentence. If a sub-
Each subject heard four stories followed ject did not remember having heard the target
by recall and then four additional stories fol- and would not venture a guess as to its mean-
lowed by recall. Half the subjects in each age ing, the experimenter read the story from the
group listened to story Set A and half listenedbeginning up to and including the proposi-
to story Set B. The presentation order of thetion after the metaphor and again asked for
stories within a set was arranged so that each the meaning (it was necessary to reread the
half of the presentation included two storiesstory on 11 occasions, or 2.8% of the time).
with metaphor targets (one at the R1 node andThe procedure for the second half of the story
one at the 01 node). A story with a literalorder was identical to that of the first. All re-
target was always followed by one with a call and probe responses were recorded on
metaphor target, and vice versa. Presentationaudiocassette tapes and later transcribed and
of the second four stories reversed the order scored.
of target types. For instance, if a particular
presentation order began with a story contain- Scoring procedure.-All the recall pro-
ing a metaphor at the Reaction node (MR), the tocols were transcribed, one story per page, as
succeeding order was: literal target (L), a sequence of propositions as recalled. Two
metaphor at the Outcome node (MO), L, independent scorers (Messe and Waggoner)
break for recall, L, MO, L, MR. Each storyscored the protocols to determine if the mean-
appeared at every possible position (first ing of the input propositions had been repro-
through eighth) in the order, yielding eight duced in recall. Subjects were judged to have
different presentation orders. These orders recalled the meaning of an input proposition
were randomly assigned to subjects within if the general idea expressed by the proposi-
each condition and grade level. tion (the "gist" of the proposition) was evi-
dent in the recall protocol. Scorers agreed
Procedure.--Subjects were tested indi- 97% of the time on whether the "gist" of a
vidually in a vacant room in the school build-
proposition was present in the recall. Dis-
ing. The testing was conducted by two of the
agreements were adjudicated by a third
authors, Messe and Waggoner. Each tested judge.
the same number of male and female children
in each grade and for each story set. Each In addition to this essentially quantitative
child was told that the experimenters were scoring of the presence or absence of the in-
interested in how well children could remem- put propositions, certain qualitative aspects of
ber stories. They were told that they would recall were scored. These qualitative scorings
listen to four short stories and then be asked were based on those defined by Mandler
to tell them back to the experimenter and that (1978) and Mandler and DeForest (1979).
this would be repeated for an additional four Since these scorings were not taken into ac-
stories. To familiarize the children with the count in the analyses, they need not be dis-
procedure, each child heard a practice story, cussed in detail. However, they did serve as
structurally identical to the eight experimen- guides in determining the scoring of
tal stories, and was asked to recall it im- metaphor recall and thus require some brief
mediately. He or she was then told to listen mention. A subject was credited with correct
very carefully to the next four stories. After recall of a proposition (metaphor or literal) if
hearing the fourth story, the subject was the recall was verbatim or considered to be an

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
1160 Child Development
gave unreasonable or literal interpretations
appropriate or slightly distorted interpretation
were not credited with having comprehended
of the input proposition, regardless of its se-
the metaphor.
quential position in the recall. Recall was not
credited when a subject failed to reproduce
any aspect of the input proposition or if he or
Results
she replaced it with a structural addition (a
proposition in the appropriate form for a Analyses will be reported for the recall of
the metaphors and their literal counterparts at
given node but consisting of fabricated or ex-
tremely distorted material) or new materialthe Reaction and Outcome nodes, recall of the
(fabricated or extremely distorted material three propositons following the metaphor and
that was not in the appropriate form of any literal propositions, probed recall of the
story node). Thus one child's recollection meaning
of of the metaphors, and, finally, the
relation between story recall of the metaphor
the metaphoric target of the "Jill at the Zoo"
story as, "First she was in a room with and
a probed recall of the meaning of the
whole bunch of kittens that were nice and metaphor.
furry," was scored as new material, for, We were particularly concerned with the
clearly, it neither captures the meaning of the
frequency with which the metaphors were re-
metaphor nor fits the form of the Reaction
called relative to their literal counterparts and
node. (Note that the child had recalled both
with whether the occurrence of a metaphor in
Settings and the Beginning; therefore this the story might have disrupted recall of the
could not be scored as a structural addition of
following propositions in the story. The first
either of these nodes.) A structural addition
analysis explored the difference in recall of
was scored when another child began her re-
metaphors and their literal counterparts in the
call of the "Jill" story with, "Well, she was a
Reaction and Outcome position of the stories.
little kitten." This proposition is in the proper
The analysis included two between-subject
form for a Setting statement, so a scoring of
factors (grade: 2, 4, 6; and story set: A, B) and
structural addition is appropriate.
two within-subject factors (proposition type:
Probe data.-From the subjects' re- metaphor, literal; and proposition position:
sponses to the questions probing for theirReaction, Outcome). The main effect of grade
was significant, F(2,90) = 9.967, p < .001, as
understanding of each metaphor, the scorers
determined whether the responses were rea-
well as the main effect of story set, F(1,90) =
sonable, literal, or unreasonable interpreta-
5.151, p < .026. The main effect of proposi-
tions of the metaphor. Agreement between tion position was also significant, F(1,90) =
scorers was obtained on 91% of the probe re-
139.616, p < .001, indicating that the proposi-
sponses. Disagreements were settled by tions
a in the Outcome position (X = 5.271)
third judge. Unlike the scoring of metaphorwere recalled much better than those in the
recall, structural appropriateness was not anReaction position (X = 2.083). The only other
issue here. Thus a response could be scored significant effect was the triple interaction of
as a reasonable interpretation and be either story set x proposition type x proposition
position, F(1,90) = 13.272, p < .001. Exami-
structurally appropriate (e.g., Reaction meta-
nation of the means in Table 1 indicates that
phor: "Jill was a kitten in a room filled with
balls of yarnm"; response: "She really liked in story set A the reaction position metaphors
it there") or structurally inappropriate (e.g.,
were not as well recalled as the literal targets
"I guess it meant she went through the zoo in the same story were, while the reverse was
like kittens play with yarn"). In the for- true in story set B. In the case of the Outcome
mer example, the subject's response was propositions,
in metaphors were better recalled
the form of a reaction, whereas in the latterthan the literal targets were in story set A, and
the response was not. Both subjects, however,the recall of metaphor and literal targets was
demonstrated an ability to interpret theabout the same in story set B. Since the
metaphor meaningfully and were thus cred- stories were counterbalanced so that stories
ited with having comprehended the meta- that had metaphors in the Reaction and Out-
phor. Responses were scored as unreason- come positions in Set A had their literal
able interpretations with appropriate struc-counterparts in Set B, this interaction seems
ture (e.g., "She was bored") or with inappro-attributable to the ease with which the Reac-
priate structure (e.g., "She always wanted totion and Outcome propositions were recalled,
see animals and stuff"). There were also regardless of their form; that is, this result is a
responses that indicated that the subject inter- characteristic of the propositional content and
preted the metaphor literally (e.g., "It would not of the literal or metaphoric nature of the
have meant she was a cat," or, "There was target proposition. If a Reaction or Outcome
some yamrn behind her"). The subjects who proposition was difficult to recall in a particu-

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
Waggoner, Messe, and Palermo 1161
TABLE 1

MEAN RECALL OF METAPHOR AND LITERAL PROPOSITIONS IN THE REACTION


AND OUTCOME POSITIONS OF STORY SETS A AND B

PROPOSITION TYPE

Literal Metaphor

Reaction Outcome Reaction Outcome

Story set A ........ .604 1.062 .375 1.292


Story set B ........ .396 1.479 .708 1.438
NOTE.-Maximum score in each cell is 2.00.

sentences, F(2,180) = 163.572, p < .001, indi-


lar story, it did not matter whether the propo-
sition was a metaphor or a literal target. cating that with increases in grade there were
increases in the likelihood of recalling the
While the above analysis seems to indi-
three sentences following both the metaphor
cate that children had no more difficulty and
re- the literal sentences combined and there
calling metaphors than they did literal propo-
were differences in recall of the three sen-
sitions, the possibility remains that the
tences following both the metaphor and the
metaphors may have required additional cog-
literal sentences. There was also a significant
nitive effort or processing capacity that inter-
interaction of story set x story type, F(1,90) =
fered with processing or encoding of subse-
4.064, p < .047, and a significant triple in-
quent propositions in the story. This
teraction of story set x story type x sentence,
possibility was examined by scoring the recall
F(2,180) = 58.135, p < .001. Table 2 presents
of the three sentences following the the mean recall of each of the three sentences
metaphors or their literal counterparts in the
following the metaphor and literal sentences
stories. Separate analyses were conducted for
in story sets A and B. Again, the interactions
the Reaction and Outcome metaphors since
result from the differential recall of the con-
the three statements following the Reaction tent of the subsets of four stories. Recall of the
position metaphors were the G1, A1, and 01
three sentences (R1, G1, and A1) in story set A
propositions, while the propositions following
following a metaphor is comparable to the re-
the Outcome metaphor were the E/B, R2, and
call of the same sentences following the lit-
G2 propositions. Both analyses involved two
eral sentences in story set B. Similarly, the
between-subject factors (grade: 2, 4, 6; and
recall of the three sentences following a
story set: A, B) and two within-subject factors
metaphor in story set B is comparable to the
(story type: metaphor, literal; and sentences:
recall of the same sentences following literal
1, 2, 3). Note that in these analyses the scores
targets in story set A. The interaction results
are for the three sentences following the from the differential recall of the three sen-
metaphor in the stories containing metaphors
tences in the different subsets of four stories
as compared to the three sentences following
and not from the effects of the literal or
the literal counterpart in the stories without a
metaphor. The results of the Reaction metaphor sentences themselves.
metaphor stories yielded a significant main ef- The same picture emerges in the out-
come metaphor stories. There was a
fect of grade, F(2,90) = 3.594, p < .032, and of

TABLE 2

MEAN RECALL OF THE THREE PROPOSITIONS FOLLOWING THE REACTION


METAPHOR IN THE Two STORY SETS

STORY TYPE

Metaphor Literal

G1 A1 01 G1 A O01
Story set A ..... .750 1.563 1.167 .292 1.167 1.833
Story set B ..... .417 1.167 1.813 .750 1.771 1.188

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
1162 Child Development

significant main effect of grade and sentence. TABLE 4


In this analysis, the story set x story type
MEAN FREQUENCY OF CORRECT INTERPRETATION
interaction was not significant, but the tripleOF THE METAPHOR MEANING FOR REACTION AND
interaction of story set x story type x sen- OUTCOME PROPOSITIONS AT EACH GRADE
tence was significant. The means in Table 3
reveal the same relationship among the Reaction Outcome
means as was seen in Table 2. No other main
Grade 2 .......... .781 1.000
effects or interactions were significant in
either of these analyses. Grade 4 .......... 1.563 1.563
Grade 6 .......... 1.594 1.688
The recall results are clear in demonstrat- Mean .......... 1.313 1.417
ing that there were grade differences in recall
NOTE.-Since there were two Reaction and two Out-
of metaphors and their literal counterparts but
come metaphors for each subject, the maximum score is
that there were no differences between recall 2.00 for each cell.
of metaphors and their literal counterparts
within any grade for either Reaction or Out-
come propositions. able interpretation of the figuratively ex-
We can now turn our attention to the re- pressed meaning of the metaphor. Since the
sults of the probe data. Table 4 presents the children were given the opportunity to give
mean frequency with which children in the two meanings if they wished, they could have
three grades were able to give a correct inter- given both a literal and figurative meaning.
pretation of the meaning of the metaphors in
This happened only once in grade 2 and
the Reaction and Outcome positions. As may
twice in grades 4 and 6. Only a few of the
be seen from those means, correct interpreta- children gave literal responses without an ap-
propriate figurative interpretation. This oc-
tion of the metaphors increased with grade,
but the interpretation of Reaction and Out- curred 22 times in grade 2 and once in grades
come metaphors differs very little, despite the
4 and 6. Thus, literal responses seldom oc-
fact that the previous analyses indicated that curred in grades 4 and 6, and on only 22 occa-
recall of the Outcome propositions was sions out of 128 did the second-grade children
significantly greater than recall of the Reac- offer a literal interpretation and no correct
tion propositions. figurative interpretation. Second graders cor-
rectly interpreted the figurative meaning on
A three way analysis of variance with two 57 occasions (see Table 5). Of the 32 second-
between-subject factors (grade: 2, 4, 6; and grade children, 28 gave at least one correct
story set: A, B) and one within-subject factor interpretation of the four metaphors pre-
(proposition type: Reaction, Outcome) was sented to them. All the children in all the
applied to the data. The main effect of grade grades gave a figurative interpretation to most
was significant, F(2,90) = 27.727, p < .001. of the metaphors, but some of those interpre-
No other main effect or interaction was tations were incorrect interpretations of the
significant. figurative meaning rather than literal interpre-
An examination of the content of the in-tations of the metaphoric proposition; thus 34
incorrect figurative interpretations were given
correct responses of the children to metaphor
probes was undertaken to determine what by second graders, 27 by fourth graders, and
22 by sixth graders.
types of interpretation errors were being
made. We were also interested in whether the A final analysis of the data was an effort
children were interpreting the metaphors lit-to relate the recall data to the probe data. At
issue was the relation between recall and
erally if they were not able to give an accept-

TABLE 3

MEAN RECALL OF THE THREE PROPOSITIONS FOLLOWING THE OUTCOME


METAPHOR IN THE TWO STORY SETS

STORY TYPE

Metaphor Literal

E/B R2 G2 E/B R2 G2
Story set A ..... 1.354 .521 .812 1.438 .667 .625
Story set B ..... 1.438 .854 .646 1.438 .542 1.000

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
Waggoner, Messe, and Palermo 1163
TABLE 5

PROBABILITY OF METAPHOR COMPREHENSION GIVEN RECALL OR FAILURE TO


RECALL THE METAPHOR

Recall Probea p Nonrecall Probe p z


Grade 2 ... 37 22 .59 91 35 .38 2.17b
Grade 4 ... 68 59 .87 60 41 .68 2.59b
Grade 6 ... 67 58 .87 61 50 .82 .778

a Number of correct interpretations of the m


b The difference between the two probabilit

knowledge of the Reaction


meaning and Outcome
of propositions
the metap in the
stories.
For each subject we More important the
tabulated than therelati
recall data
tween recall and failure to recall each however, the probe data show that, whe
metaphor in the stories and producingasked what a metaphor means, these childre
a cor-
rect interpretation of the meaning were
of the
able to give an appropriate verbal an
metaphor in the postrecall probe situation.
swer. The 7-year-old children were able to d
Table 5 presents the number of occasions so 45%forof the time, while 9- and 11-year-old
each grade when the metaphor was recalled were able to do so 78% and 82% of the time
or not recalled and the occasions when the respectively (see Table 5). It is worth noting
same metaphor was given an appropriate in- in this context that literal interpretations o
metaphors were almost never given by even
terpretation during the probe task. The proba-
the youngest children and that many of th
bility of correctly interpreting the metaphor,
given recall or nonrecall, is also indicated.interpretations
As of the meanings of the meta
may be seen in Table 5, for all three grades phors given were, while incorrect by our scor
the probability of correctly interpreting ing the criteria, figurative in nature. It seemed
metaphor in the probe task is greater afterclear
re- that the children understood at some
level
call than after failure to recall the metaphor in the figurative use of language, although
the story recall task. The difference betweenthey did not always offer the meaning we
the two probabilities is significant (p < were
.05) looking for when we scored the re-
for grades 2 and 4. In grade 6 the differencesponses.
is It seems safe to conclude from these
in the same direction, but correct interpreta-
data that, in the context of a story, children as
tion is quite high in both conditions, and young
the as 7 years comprehend and are able to
difference is not statistically significant. verbalize the meaning of metaphors in a story
as indicated by their ability both to recall and
Discussion to explain that meaning.
There are at least three issues on which
The use of a grammar for constructing the
this research bears with varying degreesstories of
used in this research provided a struc-
ture that the children could use not only to
force. First, these data provide clear evidence
aid recall
that children as young as 7 years of age can but also to allow them to succeed in
the more
and do correctly interpret metaphors that are difficult task of verbalizing the
meaning of the metaphors. In the terms used
embedded within a story schemata. Further-
by Vosniadou et-al. (1984), the story schemata
more, they are capable, under these circum-
stances, of verbally indicating what the the difficulty of the comprehension
reduced
metaphors mean. Both the recall and task the
so that it was within the child's difficulty
probe data support these conclusions. limit both to recall and to verbalize the mean-
Recall
of metaphors was the same as recallings of ofthethe metaphors. As previous research
has indicated
literal counterparts of those propositions at (e.g., Gardner et al., 1975; Sil-
verman
each of the three age levels tested. This resultet al., 1976; and Winner et al., 1976), a
was also replicated for each of the two sentence
struc-or two is not enough to bring the
tural positions in the story grammar task within the child's difficulty limit. In the
at each
Vosniadou
age level. Furthermore, recall of the three et al. study, where preschool and
first-un-
propositions following the metaphors was and third-grade children were asked to
act out
disturbed by the inclusion of a metaphor inmetaphorically expressed endings to
the story. Recall of these propositions was
stories, performance was quite high (85%) for
the youngest children when the story had a
equivalent following literal and metaphorical
highly
propositions at all grades and following both probable ending. On the other hand,

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
1164 Child Development

performance was only 23% for the same agecently, however, that, at least for short-term
group when the endings were of low proba-memory, it may reflect more than one knows
bility. Third-grade children achieved only or comprehends about a story (e.g., Omanson
68% correct responding in the latter condi- et al., 1978). In the former case, it has been
tion. While the present experiment and that of shown, for example, that inferences derived
Vosniadou et al. involve quite different proce-from the story become incorporated in one's
dures and the age groups of the two studiescomprehension of a story. In the latter case it
only partially overlap, both achieved successhas been suggested that rote retrieval of the
surface propositions of a story may occur in-
by bringing the task within the difficulty limit
of the children. It is not possible to identify dependently of comprehension of the mean-
the important variables precisely in each case, ing, at least for immediate recall of a single
but it seems likely that, by allowing the chil- story (Omanson et al., 1978). Retrieval of
dren to act out the stories as well as by plac- some representation of the surface story infor-
ing the metaphor at the end of the story, Vos- mation is certainly a prerequisite for, al-
niadou et al. may have decreased the overall though it may not be sufficient for, com-
prehension. It seems unlikely in the present
difficulty of the task and made it easier for the
children to demonstrate their knowledge. Incase, where the procedure required listening
the present experiment, the story structureto four stories prior to recalling the stories in
served the same purpose and made it possible the order of presentation, that rote recall was
for the children to verbalize the metaphorical a major factor. The amount of rote recall, how-
meaning on questioning. ever, remains an unknown. Two facts throw
some light on the issue. First, the probe data
This experiment also provides supportindicate greater comprehension of the
for the arguments advanced by Bransford and metaphors than was indicated by the recall
McCarrell (1974) and Palermo (1982, in press)data, and, second, the probability of correct
that the process of comprehending metaphor explanation of the metaphor meaning was
is the same in kind as the process of com-
considerably higher after recall than after fail-
prehending literal language. Since all the evi-
ure to recall. Both of these aspects of the data
dence is based on confirming the null hy- suggest that recall, while less sensitive than
pothesis, no strong statements can be made,specific probes for comprehension, does pro-
but it is worth noting that performance in vide a reasonable first measure of comprehen-
the recall task revealed no differences for
sion, although it is unsatisfactory as the only
metaphor and literal propositions, as would
measure. The specific explanations of the
be predicted by the theorists involved. It is meanings of the metaphors certainly provide
also the case that differences in performance more definitive answers to the comprehen-
were found between stories in recall of partic-sion question, although one may ask whether
ular Reaction and Outcome propositions. the meaning of the metaphor was com-
When either type of proposition in a particu- prehended at the time of story recall.
lar story was found to be relatively easy or The results are odd in the sense that re-
difficult in its literal form, it was also compar-
call appears to be related to comprehension
ably easy or difficult in its metaphorical form.
because the probability of correct interpreta-
Thus a proposition in a story was easy or
tion of the metaphor is much higher after re-
difficult to comprehend in the context of that
call than after failure to recall. On the other
story and not as a function of whether it was
hand, recall of reaction metaphors was much
expressed in a literal or a metaphorical form.
lower than outcome metaphors. The only evi-
Two measures of metaphor comprehen- dent account of these conflicting results
sion were used in this research: recall and comes from the observation that children re-
probed explanation of meaning. The recall spond the same way for literal statements in
measure was used because it has frequentlythe recall task. That fact suggests that children
been employed by those exploring the effects have other reasons for not including the reac-
of story grammars on story comprehension tion of the protagonist in their recall of the
(e.g., Mandler, 1978; Mandler & Johnson, story. The problem, the actions taken to solve
1977; Stein & Glenn, 1979). In all respects, the problem, and the outcome are important,
the present data confirmed the findings ofbut how the person feels is apparently consid-
those studies in showing differential recall of ered unimportant, whether that feeling is
specific propositions in the story structure. As stated literally or figuratively. It is, however,
has been well-known for some time, the recallunderstood how the protagonist feels, regard-
measure does not reflect all one knows about less of the form of expression.
a story (e.g., Brockway, Chmielewski, & Further research in this area will be
Cofer, 1974); it has been suggested more re- needed to extend these findings to younger

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
Waggoner, Messe, and Palermo 1165

ages. More important, we need to explore theG2: She wanted to go home.


nature of the relationship between the cogni- A2: She crawled out from under the lion's cage.
tive status of the child and the nature of the 02: And ran to her mother.
interpretations given to metaphors of various E: Jill decided her home was a better place to
live than the zoo.
kinds. In the present study there were clear
developmental improvements with age in References
performance on both measures. It is not at all
clear whether these changes are attributable Asch, S., & Nerlove, H. (1960). The development of
to differences in strategies in recall and expli- double function terms in children: An explor-
cation used by younger and older children, atory study. In B. Kaplan & S. Wapner (Eds.),
differences in the knowledge about the story Perspectives in psychological theory (pp. 47-
content or the knowledge base into which the 60). New York: International Universities
content of the stories is assimilated, differ- Press.
ences in interpretation of the task require-Billow, R. M. (1975). A cognitive developmental
ments, or other developmental differences in study of metaphor comprehension. Develop-
the age range examined here. mental Psychology, 11, 415-423.
Bransford, J. D., & Johnson, M. K. (1972). Contex-
Appendix tual prerequisites for understanding: Some in-
vestigations of comprehension and recall. Jour-
Two Examples of the Stories Used,
nal of Verbal Learning and Verbal Behavior,
One with the Target Metaphor in the 11, 717-726.
Reaction Position and One with the
Bransford, J. D., & McCarrell, N. S. (1974). A
Target Metaphor in the Outcome
Position sketch of a cognitive approach to comprehen-
sion: Some thoughts about understanding what
it means to comprehend. In W. B. Weimer &
"Tony and the Scratching Noise" (Metaphor
Version) D. S. Palermo (Eds.), Cognition and the sym-
bolic processes (pp. 189-229). Hillsdale, NJ:
Erlbaum.
S1: It was a cold and windy night.
S2: Tony had been asleep for hours. Brockway, J., Chmielewski, D., & Cofer, C. N.
B: Suddenly a loud scratching noise woke him (1974). Remembering prose: Productivity and
up. accuracy constraints in recognition memory.
R1: Tony was very frightened. Journal of Verbal Learning and Verbal Behav-
G1: He wanted comfort from his parents. ior, 13, 194-208.
A,: He opened his mouth to call for them.
Cometa, M. S., & Eson, M. E. (1978). Logical opera-
01: But his voice was the cry of a small mouse.'
E/B: Then he heard the noise again. tions and metaphor interpretations: A Piage-
R2: Tony became even more frightened. tian model. Child Development, 49, 649-659.
G2: He was determined to wake his parents up. Gardner, H., Kircher, M., Winner, E., & Perkins, D.
Ag: He ran to their bedroom. (1975). Children's metaphoric productions and
02: His parents told him not to be frightened. preferences. Journal of Child Language, 2,
E: Tony laughed when he found out it was only 125-141.
a branch hitting the window. Hildyard, A., & Olson, D. R. (1978). Memory and
inference in the comprehension of oral and
written discourse. Discourse Processes, 1, 91-
"Jill at the Zoo" (Metaphor Version)
118.
Mandler, J. M. (1978). A code in the node: The use
S1: Jill really loved animals.
S2: But she lived in a big city where there of a story schema in retrieval. Discourse Pro-
weren't any. cesses, 1, 14-35.
B: One day her mother took her to the zoo. Mandler, J. M. (1982). Some uses and abuses of a
R1: Jill was a kitten in a room filled with balls of story grammar. Discourse Processes, 5, 305-
yarn.2 318.
G1: She wanted to stay at the zoo for ever and Mandler, J. M., & DeForest, M. (1979). Is there
ever.
more than one way to recall a story? Child De-
Al: Jill hid under the lion's cage when it was time
to leave. velopment 50, 886-889.
O1: And her mother couldn't find her. Mandler, J. M., & Johnson, N. S. (1977). Remem-
E/B: When it got dark, the lions started making brance of things parsed: Story structure and re-
scary noises. call. Cognitive Psychology, 9, 111-151.
R2: Jill became scared. Omanson, R. C., Warren, W. H., & Trabasso, T.

1 Literal version: But his voice could hardly be heard.


2 Literal version: Jill was really excited about all the different animals.

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms
1166 Child Development

(1978). Goals, inferential comprehension, and Stein, N. L., & Glenn, C. E. (1979). An analysis of
recall of stories by children. Discourse Pro- story comprehension in elementary school
cesses, 1, 323-336. children. In R. O. Freedle (Ed.), New direc-
Palermo, D. S. (1982). Theoretical issues in seman- tions in discourse processing: Vol. 2. Advances
tic development. In S. Kuczaj (Ed.), Language in discourse processes (pp. 53-120). Norwood,
development: Syntax and semantics (pp. 335- NJ: Ablex.
364). Hillsdale, NJ: Erlbaum. Vosniadou, S., & Ortony, A. (1983). The emergence
Palermo, D. S. (in press). From the marble mass of of the literal-metaphorical-anomolous distinc-
language, a view of the developing mind. tion in young children. Child Development, 54,
Metaphor. 154-161.

Reynolds, R. E., & Ortony, A. (1980). Some issues Vosniadou, S., Ortony, A., Reynolds, R. E., & Wil-
in the measurement of children's comprehen- son, P. T. (1984). Sources of difficulty in the
sion of metaphorical language. Child Develop- young child's understanding of metaphorical
ment, 51, 1110-1119. language. Child Development, 55, 1588-
Rumelhart, D. E. (1975). Notes on a schema for 1606.
stories. In D. G. Bobrow & A. Collins (Eds.), Whaley, J. (1981). Readers' expectations of story
Representation and understanding: Studies in structures. Reading Research Quarterly, 17,
cognitive science (pp. 211-236). New York: 90-114.
Academic Press. Winner, E., Engel, M., & Gardner, H. (1980). Mis-
Silverman, J., Winner, E., & Gardner, H. (1976). On understanding metaphor: What's the problem?
going beyond the literal: The development of Journal of Experimental Child Psychology, 30,
sensitivity to artistic symbols. Semiotica, 18, 22-32.
291-312. Winner, E., Rosenstiel, A. K., & Gardner, H. (1976).
Stein, N. L. (1982). What's in a story: InterpretingThe development of metaphoric understand-
ing. Developmental Psychology, 12, 289-
the interpretations of story grammars. Dis-
course Processes, 5, 319-335. 297.

This content downloaded from 202.43.93.14 on Sat, 30 Sep 2017 03:46:29 UTC
All use subject to http://about.jstor.org/terms

You might also like