You are on page 1of 8

Metaphor and Symbol

ISSN: 1092-6488 (Print) 1532-7868 (Online) Journal homepage: https://www.tandfonline.com/loi/hmet20

Structure-Mapping and the Simile -- Metaphor


Preference

Ravid A. Aisenman

To cite this article: Ravid A. Aisenman (1999) Structure-Mapping and the Simile -- Metaphor
Preference, Metaphor and Symbol, 14:1, 45-51, DOI: 10.1207/s15327868ms1401_5

To link to this article: https://doi.org/10.1207/s15327868ms1401_5

Published online: 17 Nov 2009.

Submit your article to this journal

Article views: 220

View related articles

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=hmet20
METAPHOR AND SYMBOL, 14(1),45-51
Copyright 0 1999, Lawrence Erlbaum Associates, Inc.

Structure-Mapping and the


Simile-Metaphor Preference
Ravid A. Aisenman
Department of Linguistics
Tel Aviv University

This article aims to provide cognitively motivated evidence for the nonequivalence
view of simile and metaphor. This evidence was derived from a study based on two
preference tasks in which participants were asked to select the linguistic form that
conveys the comparison most adequately and naturally. Adopting Gentner's (1983)
"Structure-Mapping'' analysis, I was able to establish a correlation between the type
of comparison (attributive, relational, or double) and the means speakers select to best
represent each such comparison.The results of this study strongly suggest that similes
are the preferred linguistic representation for mapping attributive predicates, whereas
metaphors are favored for mapping relational predicates. The study further consid-
ered the results of the double comparisons.

Research on figurative language concerned with differences between metaphor


and simile has focused on the semantics of the two structures or on how they are
understood. Relatively little interest has been shown in the functional difference
between these two structures. This article concerns these two linguistic struc-
tures found in both poetic and nonpoetic contexts: metaphor ("A is B") and sim-
ile ("A is like B").
The study analyzed one aspect of the difference between these two construc-
tions and specified the conditions under which speakers prefer one or the other.
Analysis focuses on the effect of the shared predicate type on linguistic structure
preference and is based on evidence provided by preference tasks that reveal a
clear distribution.

Requests forreprintsshould be sent to Ravid A. Aisenman,Department of Linguistics,Tel Aviv Uni-


versity, Ramat Aviv, 69978, Tel Aviv, Israel. E-mail: ravid@post.tau.ac.il
There are two major views of metaphor and simile: the equivalence view and
the nonequivalence view. For many years, leading theories of metaphor, following
the Aristotelian tradition, regarded these figures of speech as equivalent. This is
also true of the more current conceptual approach following Lakoff and Johnson
(1980). Under this approach, metaphor is not a linguistic phenomenon but rather a
conceptual phenomenon consisting of a mapping process between two do-
mains-the base and the target. This entails rejecting any difference between simi-
les and metaphors because they differ linguistically only in the surface presence or
absence of the word like. In this conceptual view, these figures are treated the same
with regard to comprehension, interpretation, and usage.
Several psycholinguistic studies have challenged this claim. They analyze the
nonequivalent aspects of similes and metaphors as two essentially different con-
structs, each involving distinct processes of comprehension and different condi-
tions of usage. Such attempts have considered inter alia the different
interpretations assigned to these structures (Verbrugge, 1980) and the role of ab-
stractness in structural preference (Gibb &Wales, 1990), as well as certain devel-
opmental trends (Reynolds & Ortony, 1980).
The occurrence or absence of the word like is not merely an insignificant lin-
guistic difference; rather, these two structures reflect and impinge on different
cognitive processes and are used for different functions. This is forcefully ar-
gued in the framework of the "class-inclusion" theory of Glucksberg and Keysar
(1990). Taking this as a starting point, I investigate one as yet unexplored aspect
of the nonequivalence of similes and metaphors: the connection between the
shared predicate type and speakers' preferences when choosing between a simile
and a metaphor.
Because this analysis takes into account the type of predicates shared by the
base and target, it calls for a theory of metaphor that makes an explicit distinction
between types of predicates. This kind of theory is Gentner's (1983) well-known
"Structure-Mapping." Structure-mapping theory focuses on the interpretation
rules of analogy and metaphor. The interpretation process involves mapping predi-
cates from one conceptual domain (the base) onto the other conceptual domain
(the target). Gentner's main interest was to define the predicates relevant to this
mapping process to predict which predicate will be mapped, and hence what the
metaphor's interpretation will be. The essence of Gentner's structure-mapping
theory is the distinction between relational and attributive predicates. Simply
stated, attributive predicates are one-place predicates (e.g., "X is straight"),
whereas relational predicates are n-place predicates (e.g., "X is used to transfer Y
to Z"). Structure-mapping's interpretation rules state that attributes of the object
that occur in the base domain are discarded and so do not get mapped, whereas re-
lational predicates of the base are preserved and carried across to the target.
In a later study, Gentner and Clement (1988) used the attributive-relational dis-
tinction to derive a further distinction between three types of comparison:
SIMILE-METAPHOR PREFERENCE 47

Attributive comparisons-only attributive predicates are mapped (e.g.,


"the sun is like an orange ";the mapped predicates are: "round, " "orange ").
Relational comparisons--only relational predicates are mapped (e.g., "a
sea is like a big aquarium "; the mapped predicate is: "X contains Y").
Double comparisons-here the objects share both attributive and rela-
tional predicates, but the relational mapping is preferred (as in "a stem is like
a drinking straw ").

Gentner and Clement (1988) loosely named all these comparisons "meta-
phors," although all the material they used was in fact in the form of similes. This
inconsistency clearly reflects their conviction that the linguistic structure plays a
negligible role during the interpretation process.
In marked contrast to this approach, the study reported here used the attribu-
tive-relational distinction to describe speakers' structural preferences between
metaphor and simile. It is argued that there is a correlation between the type of
predicate mapping employed by the two compared objects and the type of lin-
guistic structure chosen to represent this mapping. The study further proposed
that the simile form is preferred to represent mapping of attributive predicates,
whereas the metaphor is favored in the case of relational predicates mapping.
Double comparisons will probably distribute more or less evenly between the
two forms.

STUDY 1

Twenty-three participants, all native speakers of Hebrew, took part in the study. A
two-part questionnaire was administered. The first part included 19 pairs of state-
ments, each with two concrete objects inserted into the two different structures, the
first a metaphor (e.g., "the sun is an orange") and the second a simile (e.g., "the sun
is like an orange"). Based on responses from a pilot study of eight He-
brew-speaking judges, 13 out of the 19 items were taken from Gentner and Clement
(1988) and combined with 6 new items. Participants were asked which of the two
comparisons was more natural or suitable. In the second part, participants were
asked to list the most salient common features of each object pair in isolation. Ac-
cording to the features listed by the participants (in the second part), each compari-
son was classified as attributive, relational, or double (see Appendix).
The findings in Table 1 indicate that speakers clearly prefer to represent attribu-
tive comparisons in simile form (85.5%) and that they tend to prefer to represent
relationals in metaphor form (57%).The significant role of relational predicates is
revealed by the increased preference for metaphors in double (31.5%) versus at-
tributive comparisons (14.5%). This increase can only be attributed to the addi-
tional relational predicates in double comparisons. The results are statistically
48 AISENMAN

TABLE 1
Distribution in Percentage of Similes and Metaphors
in the Different Comparison Type (N= 23)

Simile Metaphor

Attributive
Relational
Double

significant,~ 2 ( 2N, = 23) = 70.05, p < 0.0000, and support the argument for a corre-
lation between the type of predicate mapped and the means by which the language
best represents this comparison.
This distribution shows the general preference for similes predicted by the
findings of Gibb and Wales (1990). They considered the role of the relative con-
creteness or abstractness of the base and target element in speakers' preference
for simile or metaphor, respectively, and found that similes are clearly preferred
when both base and target are concrete. Because the experiment reported here
included only concrete objects, this general preference was expected. However,
the concrete objects that shared relational features were still preferred in the
metaphor form.

STUDY 2

In the second experiment,specific predicates were activated and mapped to directly


relate between the mapping process and the linguistic form. For this purpose, only
the attributive and relational comparisons were taken-six comparisons of each
type. The statements appeared in pairs of metaphor, followed by a simile as in
Study 1, with one additional factor-specification of the salient common feature
shared by the two objects (e.g., "the sun is (like) an orangev-both are orange).
The assumption was that having the salient common feature explicitly stated would
force participants to perform a mapping on it and this would be reflected in their
structural preference. Twenty participants were included, and the instructions were
the same as in Study 1.
Table 2 shows that the second experiment yielded similar results to the first
one: 90% of the attributive comparisons were preferred in simile form, and 54%
of the relational comparisons were favored in metaphor form. The results are
, = 20) = 51, p < 0.0000, even though the
again statistically significant, ~ 2 ( 1N
expected increase in the metaphor-relational correlation is not found. This sug-
gests that the mapping process is an individual decision that cannot be directed
or forced.
SIMILE-METAPHOR PREFERENCE 49

TABLE 2
Distribution in Percentage of Similes and Metaphors
in the Different Comparison Type (N= 20)
- -

Simile

Attributive
Relational

GENERAL DISCUSSION

The correlation revealed between similes and attributive comparisons, metaphors


and relational comparisons can be explained by the combined factors of the func-
tion of each structure and by the nature of each predicate. The use of similes and
metaphors is not accidental. Speakers make their choices in full awareness of the
distinct function of each construction. They realize that communicating resem-
blance through metaphor, which is in fact an assertion of identity, commits one far
more than using a simile that contains the hedging term like and so is more restricted
in the degree of similarity it can suggest. Glucksberg and Keysar (1990) suggested
that the different communicativefunction of metaphor and simile stems from either
a whole or partial attribution of features. Their approach thus regards the difference
in quantitative terms of how many features can be mapped.
The study reported here instead proposed a qualitative explanation for this dif-
ference. It suggests that in the process of choosing a linguistic form, speakers are
guided mainly by the type of predicate mapping involved in each case. Whereas re-
lational features typically refer to the function of the object (a fact that is closely re-
lated to the existence of a second argument), attributive features are usually
associated with the form or structure of the object. It could well be that because
function is more crucial to an object than its form, speakers will more readily rep-
resent a functional resemblance by using the stronger linguistic tool-the meta-
phor. The attributive features, in contrast, which describe what Gentner (1983)
called "mere appearance," can be adequately expressed by a weaker linguistic rep-
resentation of resemblance-namely, the simile.
These claims for "superiority" of relationals find support in Gentner and Clem-
ent (1988), who provided strong empirical evidence for the selectivity of relational
predicates during the interpretation process. They revealed two important aspects
of this superiority. The first is that a comparison interpretation is rated higher in
relationality than in attributionality. This suggests that relational predicates are
more likely to survive the processes of mapping and interpretation, indicating the
relative strength and dominance of relational features. Another piece of evidence
for relational selectivity is the high correlation between the degree of relationality
of interpretation and the degree of aptness of comparison.
50 AISENMAN

Gentner and Clement (1988) suggested that this might show that the rela-
tional structure serves as a selection constraint tacitly applied in the interpreta-
tion process. If so, this would strengthen the argument in favor of the
essentiality of relational features. Moreover, if comparison interpretations reveal
such relational selectivity, it is not surprising that this selectivity is also evident
in a structural preference task. The study reported here can best be regarded as
an attempt to extend the implications of Gentner's (1983) original distinctions
and has important implications for the reality of structure-mapping. These dis-
tinctions seem relevant for explaining the difference between speakers' use of
similes compared with metaphors.
In conclusion, this study reinforces and provides empirical evidence for the
nonequivalence view proposed earlier. It shows very clearly that the conceptual
distinction between attributes and relations does affect the way in which speakers
use their language. Hence, we cannot analyze metaphors solely on a conceptual
level while disregarding their linguistic realization.

ACKNOWLEDGMENTS

This article was presented at the Conference on Researching and Applying Meta-
phor 11, a Seminar of the Danish Network for Metaphor, Culture, and Cognition,
May 29-3 1, 1997, Copenhagen.
I am grateful to Yeshayahu Shen and Ruth Berman for their support in develop-
ing these ideas and for their helpful comments on earlier drafts.

REFERENCES

Gentner, D. (1983). Structure-mapping: A theoretical framework for analogy. Cognitive Science, 7,


155-170.
Gentner, D., & Clement, C. (1988). Evidence for relational selectivity in the interpretation of analogy
and metaphor. In G. H. Bower (Ed.), The psychology of learning and motivation: Advances in re-
search and theory (Vol. 22, pp. 307-358). New York: Academic.
Gibb. H., &Wales, R. (1990). Metaphor or simile: Psychological determinants of the differential use of
each sentence form. Metaphor and Symbolic Activity, 5, 199-21 3.
Glucksberg, S., & Keysar. B. (1990). Understanding comparisons: Beyond similarity. Psychological
Review, 97, 3-18.
Lakoff, G.. &Johnson, M. (1980). Metaphors we live by. Chicago: University of Chicago Press.
Reynolds, R. E., & Ortony, A. (1980). Some issues in the measurement of children's comprehension of
metaphorical language. Child Development, 51, 1 11G1119.
Verbrugge, R. R. (1980). Transformation in knowing: A realist view of metaphor. In R. P. Honeck & R.
R. Hoffman (Eds.), Cognition and figurative language (pp. 87-125). Hillsdale, NJ: Lawrence
Erlbaum Associates. Inc.
SIMILE-METAPHOR PREFERENCE 51

APPENDIX

TABLE A1
Items Used in Study 1 With Percentage of Preference for Each Pair
- -

Objects Compared Simile Form Metaphor Form

ladder-hill
cloud-cottona
kite-bird
football-egg
cloud-sponge
moon-ping pong balla
roof-hats
sun-orangea
plant stem-drinking straw
television-magnets
snake-hosea
lake-mirrolq
sea-big aquariuma
soap suds-whipped creama
grass-hair
Reagan-shark
window-eyea
tiger-zebraa
eyelidxurtaina

altems included in Study 2 .

You might also like