Professional Documents
Culture Documents
INT3 Marscharketal1983.Dimensionsofmetaphor
INT3 Marscharketal1983.Dimensionsofmetaphor
net/publication/226853341
Dimensions of metaphor
CITATIONS READS
65 250
3 authors, including:
Albert N. Katz
The University of Western Ontario
105 PUBLICATIONS 2,448 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Albert N. Katz on 29 October 2015.
Dimensions of Metaphor
Two rating studies examined several dimensions of metaphorical sentences. A pool oj260
metaphors was constructed, all in the form "(noun phrase) is~are (noun phrase)." In Study 1
all of the items, and in Study 2, 98 of the items were evaluated on ten scales presumed to be
important to the comprehension or interpretation of metaphors: semantic relatedness of the
subject and predicate, comprehensibility, irnageability, imageability of the subject (topic),
imageability of the predicate (vehicle), degree of metaphoricity, metaphor goodness, ease
of interpretation, number of alternative interpretations, and felt familiarity of the meta-
phoric ground. Both experiments revealed the rated dimensions to be highly interrelated,
but some analyses allowed evaluation of alternative predictions based on current theoretical
approaches to metaphor quality and interpretation. The results indicated consistent but
mixed support for the general poisitions under consideration as each appeared to have
strong and weak areas of applicability. The interrelationships among the scales are
discussed, together with implications of the findings for current theories and future
metaphor research.
INTRODUCTION
Johnson (1976; Johnson & Malgady, 1981), for example, proposed that
perceptual mechanisms analogous to those postulated for Gestalt prin-
ciples (cf. Asch, 1958) underlie metaphoric processing. One implication of
this is that since a good metaphor is one that lends itself to a single
interpretation or Gestalt (Malgady & Johnson, 1976, p. 51), metaphor
goodness should be inversely related to the number of alternative interp-
retations available.
The attention given imagery in recent language research also suggests
its importance to psychological studies of metaphor. For the present,
primary interest lay in the relationship of imagery to several interpretability
measures. In addition to overall metaphor imagery, subject and predicate
imageability were evaluated separately in this study. These scales were
included on the basis of a suggestion made by Paivio (1979) that vehicle
imagery should be more important than topic imagery in the interpretation
of metaphors. His proposal was based on the assumption that, "the vehicle
serves as an efficient conceptual peg for metaphor comprehension to the
extent that it promotes retrieval of images and verbal information that
intersects with information aroused by the topic" (cf. Verbrugge &
McCarrell, 1977, p. 168). This suggests that the importance of vehicle
imagery for interpretability may vary with the semantic relatedness of the
corresponding topic.
One further language variable of interest here was felt familiarity. All
of the metaphors in the present study were novel to the extent that they
were either created by the authors or drawn from recent research in the
area presumed to be unknown to our introductory psychology subjects.
Nonetheless, the idea expressed by any particular metaphor, even if novel,
may be relatively familiar since some uses of metaphor are more common
than others. Metaphors likening love or a loved one to flowers or heavenly
bodies ("Juliet is the sun"), for example, are more common than those
likening history to a magnet or a mountain range to headstones. While this
variable has yet to be empirically studied with specific regard to meta-
phors, the importance of the familiarity variable in the verbal learning
tradion argued for its inclusion here.
In summary, the purpose of the present study was to determine and
evaluate possible relationships among several variables currently deemed
important in metaphor research. Since that research involves a variety of
theoretical orientations and empirical traditions, several alternative pre-
dictions were possible. These were not all definitive nor mutually exclusive
but did provide bases for evaluating several current models of metaphor
processing despite the vagueness of their current formulations. The expec-
tations from the anomaly perspective were that greater topic-vehicle re-
latedness should increase a metaphor's ease of interpretation, compre-
Dimensions of Metaphor 21
GENERAL METHOD
STUDY 1
Procedure and subjects. Each subject received a test booklet con-
taining one of 10 instruction sheets and 319 to-be-rated sentences. These
consisted of the 260 metaphors plus 59 item repetitions intended for use in
reliability checks. Twenty-nine of the latter were randomly distributed
throughout the booklet: the other 30 were first presented in a block
halfway through the list and were repeated at the end of the booklet. Each
of the ten variable scales was rated by a different group of approximately
33 subjects (range 32 to 35). All 334 subjects were University of Western
Ontario undergraduates, participating to satisfy part of a course require-
ment.
Test materials were randomly assigned in all testing sessions. Sub-
jects read the instruction sheet and were self-paced in their ratings of the
items. They were told to rate each item on its own, even if they had
encountered it before (as a reliability item).
Reliability of Measttres
Reliabilities of each of the 10 variables was assessed both within and
between subjects. Within-subject reliability measures were obtained by
calculating Pearson correlations between the first and second occurrence
of each of the 59 reliability items. These correlations were generally quite
high; at least 75% of the coefficients for each variable were significant at
or beyond the .001 level. The median coefficients and the number of
reliable correlations (out of 59) are shown in Table I. There were no
particular items that were consistently unreliable. Between-subject reli-
ability was assessed by computing Cronbach alpha coefficients for the
Dimensions of Metaphor 23
ratings o f each of the ten variables. All of the coefficients obtained were
greater than .95 (see Table I) indicating surprisingly high reliability across
o u r normative sample.
M e a n ratings (and standard deviations) were calculated for each of
the 260 metaphors in the pool. In order to eliminate any artifactual
familiarity effects, only the first rating of each of the repeated items was
included in subsequent analyses. Nevertheless, repetition effects were
e x a m i n e d by performing t-tests between the ratings for first and second
o c c u r r e n c e o f each of the 29 randomly distributed reliability items. Out of
the 290 tests (29 for each of ten scales), only 31 (11%) yielded significant
differences at or beyond the .05 level. These were fairly evenly dis-
tributed across all ten variables. Thirteen of the significant effects
reflected higher ratings for first occurrences and 18 reflected higher
ratings for second occurrences of the reliability items. This difference was
not significant by a sign test. Similar tests for the 30 blocked repetitions
could not be obtained because, contrary to instructions, approximately
25% of the subjects failed to rate the second set of these items.
R e l a t i o n s h i p s A m o n g the Ten M e a s u r e s
Table II. Intercorrelations Among the Ten Metaphor Scales (Study 1)"'/~
i i i i
1 2 3 4 5 6 7 8 9 10
l. Imageability (/) - .90 .90 .90 .61 .80 .86 .79 .67 .91
2. Subject I - .86 .86 .62 .79 ,85 .78 .62 .90
3. Predicate I - .87 .54 .72 .81 .76 .61 .85
4. Ease of interpretation - ,65 .80 .87 .82 .65 .93
5. Degree of
metraphoricity - .63 .74 .65 .36 ,67
6. Felt familiarity - .82 ,78 .69 .82
7. Semantic relatedness - ,88 .60 .88
8. Metaphor goodness - .58 .82
9. Number of alternatives - .65
10. Comprehensibility
adJ = 258.
bAll p's < .01.
Multiple Correlations
In order to clarify the relationships among the dimensions of interest,
ten multiple regression analyses were performed, one for each measure.
In e v e r y case, one of the measures was treated as a criterion variable
while the other variables were entered as predictors. A stepwise procedure
was e m p l o y e d such that the predictors maintained in the final regression
e q u a t i o n were those, and only those, which added significant predict-
ability (~ = .05) even when the influence of all the other variables was
partialled out. The ten regression equations are outlined in Table III.
Reading across the Table are the criterion variable, the multiple R
obtained, the constant (X) added to the equation which maximized
prediction o f the criterial raw score, and the predictor variables which had
a reliable and statistically independent relationship with the criterion.
Dimensions of Metaphor 25
e~
t_
I I
t~
Z ~2~
26 Marschark, Katz, and Paivio
Thus, reading across the first row, one can predict the raw score obtained
on the degree of metaphoricity ratings by the formula, R = .74 (semantic
relatedness score) + .33 (comprehensibility) - .26 (predicate imagery)
- . 14 (no. of alternatives) + 1.41; the multiple R = .76. These results
indicated that sentences were more likely to be considered metaphorical if
they were of low semantic relatedness, difficult to comprehend, had a
predicate (i.e., vehicle) that was easy to image, and was a sentence with
many alternative interpretations. The other nine regression equations can
be interpreted in a similar fashion.
Theoretical Considerations
STUDY 2
Table IV. Mean Scale Ratings (and Standard Deviations) for 260-Item Stimulus
Pool and the 98-Item Subset of Study 2, as Obtained in Study 1, and New Mean
Ratings Obtained for 98-|terns in Study 2
m
Study 1 Study 2
Method
Stimuli and procedure. Mean scale ratings for the 98-item subset
used in Study 2 were calculated from the data of Study 1. These are
p r e s e n t e d in Table IV where it can be seen that both the means and
standard deviations for all of the scales were somewhat higher than those
o f the complete 260-item pool. The differences in the means were all
relatively small (between .07 and .35), however, and appeared unlikely to
affect the results.
The p r o c e d u r e was identical to that of Study 1. The 98 metaphors
were randomly ordered on four pages of a test booklet and page order was
r a n d o m across both subjects and rating scales. Each booklet contained a
single instruction page; instructions were identical to those used in
S t u d y 1.
Subjects. The subjects were all undergraduate students at the Uni-
veristy of N o r t h Carolina at Greensboro, participating to satisfy part of a
c o u r s e requirement. Ratings on each of the ten scales were obtained from
a different group of approximately 30 subjects. A total of 303 subjects
w e r e tested in groups of 3 to 34.
30 Marschark, Katz, and Paivio
Reliability of Measures
Pearson correlation
Alpha coefficient between
coefficient Studies 1 and 2a
Table VI. Intercorrelations Among the Ten Metaphor Scales (Study 2)~`'t~
1 2 3 4 5 6 7 8 9 10
1. Imagability(/) - .89 .93 .94 .67 .88 .91 .85 .24 .93
2. Subject I - .84 .84 .52 .75 .78 .68 .17" .82
3. Predicatel - .91 .61 .89 ,86 .86 .32 .89
4. Easeofinterpretation - .71 .90 .93 .89 .25 .95
5. Degreeofmetaphoricity - .69 .82 .62 .01"* .69
6. Felt familiarity - .87 .83 .41 .91
7. Semantic relatedness - .86 .15" .91
8. Metaphor goodness - .34 .87
9. Number of alternatives - .23
10. Comprehensibility
adf = 96.
bAllp's -< ,01 except: *p < .07 (2); **N.S. (1).
C a n a d i a n s t u d e n t s o f S t u d y 1 a n d the A m e r i c a n (mostly s o u t h e r n )
s t u d e n t s o f S t u d y 2. It is u n c l e a r , h o w e v e r , w h y such differences w o u l d
h a v e a f f e c t e d o n l y o n e o f the rating scales.
Multiple Correlations
T h e s a m e s t e p w i s e p r o c e d u r e a n d criteria e m p l o y e d in S t u d y 1 were
u s e d h e r e . T h e t e n r e g r e s s i o n e q u a t i o n s that r e s u l t e d are p r e s e n t e d in
32 Marschark, Katz, and Paivio
Table VI. Those variables that were also significant predicators in the first
study are underlined to indicate their interpretative importance.
Despite restrictions in selecting 98 items for the second experiment
from the original 260, the overlap between the two studies was impres-
sive. Fully 27 of the 39 significant predicators found in the multiple
regressions of Study 1 also emerged in Study 2. A 2 x 2 chi-square
analysis (lst study predictor/no predictor vs 2nd study predictor/no
predictor), in which the frequency data were the number of correlations
that fell into each of the 4 cells, confirmed the dependence of the two
studies X2 = 6.52, p < .025. Moreover of the 12 failures, i.e., predictors
found in Study 1 but not significant in Study 2, 6 were marginally
significant.
Theoretical Considerations
~t~.1~
I I"
O
L~
I"
II Iii
34 Marschark, Katz, and Paivio
GENERAL DISCUSSION
Langer, 1948; Miller, 1979; Paivio, 1971, Ch. 13, 1979; Verbrugge, 1977).
Paivio (1979) discussed the contributions of both imagery and verbal
processes (as separate but interconnected systems) in metaphor com-
prehension. He emphasized the organizational characteristics of the
imagery system in providing a large storehouse of potentially relevant,
detail-rich information. At the same time, verbal information increases
the probability of finding a connection between the topic and vehicle of
the metaphor and keeps the search process "on track." Paivio further
suggested that the vehicle is the more potent term of a metaphor insofar as
it is its properties, by definition, that are transferred to the topic. He
therefore concluded that imagery of the vehicle should be crucial to
imaginal interpretation of a metaphor because a concrete term provides
for more rapid access to the information-rich images.
The data obtained in Study 1 supported Paivio's analysis inasmuch as
sentences rated high in metaphoricity and comprehensibility tended to
have predicates (vehicles) that were easy to image (see Table II). In Study
2, the simple correlational data supported this view, although the multiple
correlation data on the restricted stimulus subset did not. These rela-
tionships are consistent with imagery playing some role in metaphor
comprehension and at the least contradict the claim of Reichmann and
Coste (1980) that it is a "tangential phenomenon."
While imaginal processes are doubtless important in the com-
prehension of those metaphors that involve comparisons of perceptual
attributes, we would not wish to argue that visual imagery is necessary for
the interpretation of all metaphors. In particular, while research in related
areas has indicated imagery to aid in the processing of some abstract
attributes, like the pleasantness of an object (Paivio & Marschark, 1980),
whether or not it is important for the comprehension of abstract meta-
phors remains an empirical question (cf. Marschark & Paivio, 1977).
Further, the effects of imagery in sense modalities other than vision have
not really been addressed. The elimination of auditory-related metaphors
in the second experiment, however, appears to have had some effect on
the resulting pool, suggesting the need for further investigations in this
regard.
Several additional points should be mentioned in conclusion. First,
while extensive by curreht standards in the area in terms of the number of
metaphors employed and participants tested, the present data are based
only on intuitively-generated metaphors. The relationships observed
among the ten measures employed in the present study may not be the
same as those found to be relevant to metaphors generated in other ways
(e.g., Johnson, 1975; Reyna, Note 4). The similarity of the relationships
38 Marschark, Katz, and Paivio
REFERENCE NOTES
REFERENCES
Anderson, C. C. The psychology of the metaphor. The Journal of Genetic Psychology, 1964,
105, 53-73.
Asch, S. E. The metaphor: A psychological inquiry. In R. Tagiuri & L. PetrnUo (Eds.),
Person perception and interpersonal behavior. Standford, California: Stanford Uni-
versity Press, 1958.
Billow, R. Metaphor: A review of the psychological literature. Psychological Bulltin, 1977,
84, 81-92.
Campbell, P. Metaphor and linguistic theory. Quarterly Journal of Speech, 1975, 61, 1-12.
Glucksberg, S., Gildea, P., & Bookin, H. G. On understanding nonliteral speech: Can
people ignore metaphors? Journal of Verbal Learning and Verbal Behavior, 1982, 21,
85-98.
Dimensions of Metaphor 39
Honeck, R., & Hoffman, R. (Eds.), Cognition and figurative language. Hitlsdale, New
Jersey: Erlbaum, 1980.
Honeck, R., Reichmann, P., & Hoffrnan, R. Semantic memory for metaphor: The
conceptual base hypothesis. Memory and Cognition, 1975, 3, 409-415.
Johnson, M. Some psychological implications of language flexibility. Behaviorism, 1975, 3,
87-95.
Johnson, M., & Malgady, R. Some cognitive aspects of figurative language: Association and
metaphor. Journal of Psycholinguistic Research, 1979, 8, 253-265.
Johnson, M. E., & Malgady, R. Toward a perceptual theory of metaphoric comprehension.
In R. Honeck & R. Hoffman (Eds.), Cognition and figurative language. Hillsdale, New
Jersey: Erlbaum, 1980.
Katz, J. J. Semi-sentences. In J. Fodor & J. Katz (Eds.), The structure of language.
Englewood Cliffs, New Jersey: Prentice-Hall, 1964.
Katz, J. J. Semantic thory. New York: Harper and Row, 1972.
Koen, F. An intra-verbal explication of the nature of metaphor. Journal of Verbal Learning
and Verbal Behavior, 1965, 4, 129-133.
Langer, S. Philosophy in a new key. Cambridge: Harvard University Press (Reprinted by
Mentor Brooks, New York, 1948).
Malgady, R. G., & Johnson, M. G. Modifiers in metaphors: Effects of constituent phrase
similarity on interpretation of figurative sentences. Journal of Psyehofnguistic Re-
search, 1976, 5, 43-52.
Malgady, R., & Johnson, M. Measurement of figurative language: Semantic feature models
of comprehension and appreciation. In R. Honeck & R. Hoffman (Eds.), Cognition and
figurative language. Hillsdale, New Jersey: Erlbaum, 1980.
Marschark, M., & Paivio, A. Integrative processing of concrete and abstract sentences.
Journal of Verbal Learning and Verbal Behavior, 1977, 16, 217-232.
Miller, G. Images and models, similies and metaphors. In A. Ortony (Ed.), Metaphor and
thought. Cambridge: Cambridge University Press, 1979.
O'Neill, B., & Paivio, A. Semantic constraints in encoding judgments and free recall of
concrete and abstract sentences. Canadian Journal of Psychology, 1978, 32, 3-18.
Ortony, A. Metaphor and thought. Cambridge: Cambridge University Press, 1979.
Paivio, A. Imagery and verbal process. New York: Holt, Rinehart, & Winston, 1971
(Reprinted Hillsdale, New Jersey: Erlbaum, 1971).
Paivio, A. Psychological processes in the comprehension of metaphor. In A. Ortony (Ed.),
Metaphor and thought. Cambridge: Cambridge University Press, 1979.
Paivio, A., & Marschark, M. Comparative judgments of animal intelligence and pleasant-
ness. Memory and Cognition, 1980, 8, 39-48.
Pollio, H., & Burns, B. The anomaly of anomaly. Journal of Psycholinguistic Research,
1977, 6, 247-260.
Reichmann, P., & Coste, E. Mental imagery and the comprehension of figurative language:
Is there a relationship? In R. Honeck & R. Hoffman (Eds.), Cognition and figurative
language. Hillsdale, New Jersey: Erlbaum, 1980.
Skinner, B. F. Verbal behavior. New York: Appleton, Century, Crofts, 1957.
Tourangeau, R., & Sternberg, R. J. Aptness in metaphor. Cognitive Psychology, 1981, 13,
27-55.
Verbrugge, R. Resemblances in language and perception. In R. E. Shaw & J. D. Bransford
(Eds.), Perceiving, acting, and knowing: Toward an ecological psyshology. Hillsdale,
New Jersey: Erlbaum, 1977.
40 Marschark, Katz, and Paivio
APPENDIX