On the Structure and Understanding of Poetic Oxymoron

Author(s): Yeshayahu Shen

Source: Poetics Today , 1987, Vol. 8, No. 1 (1987), pp. 105-122
Published by: Duke University Press

Stable URL: https://www.jstor.org/stable/1773004

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 [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Duke University Press is collaborating with JSTOR to digitize, preserve and extend access to
Poetics Today

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 ON THE STRUCTURE AND
UNDERSTANDING OF
POETIC OXYMORON*

YESHAYAHU SHEN
Poetics and Comparative Literature, Tel Av

1. INTRODUCTION

One of the main thrusts of theories of poetic languag

the early works of the Russian formalists, has been th
tempt to draw a distinguishing line between poetic an
poetic language, that is, to define those characteristics
ic language that make it "poetic," as opposed to "non-p
It would seem reasonable to assume that theories of
tive language whose main concern is the investigation
ures of speech, try to distinguish between the poetic a
poetic: between poetic and non-poetic metaphors, poe
non-poetic oxymora etc. The fact is, however, that th
tion is relatively rarely addressed within theories of
language. Their main concern has been the discussion of
problems such as the definition of figurative language and
particular instances of figures of speech (metaphor, simile
etc.) and their interpretation (cf. Black 1962 and Beardseley
1958, whose work makes up most of the very many books and
papers on the subject). Among those which have addressed
the poetic/non-poetic distinction, the most extensive treatment
has been dedicated to the theory of metaphor. Generally, a
distinction can be made between two different approaches to
the distinction between poetic and non-poetic metaphors. The
first assumes that the distinction between poetic and non-poet-
ic metaphors is based on the criterion of "petrification."

*Many of the ideas elaborated in the present paper (in particular those presented in Sec-
tions 2 and 3) are drawn from Reuven Tzur's writings (in particular, Tzur [1983]), and
from participation in a workshop on "Cognitive Poetics," Tel Aviv University, 1983 and
1984. Thanks are also due to Rachel Giora and Ruth Ronen for their helpful comments
on a preceding draft of this paper.

Poetics Today, Vol. 8:1 (1987) 105-122

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
106 YESHAYAHU SHEN

Whereas the petrified or "dead" metaphors are u

lent in non-poetic texts, it is rather the "living" me
is prevalent in poetic texts.
The second approach relies on the distinction b
ous "understanding procedures" which are us
comprehension of metaphors. Thus, for exam
(1976) claims that there are unique procedures w
tute what he calls "Poetic Competence," for the
sion of poetic texts. These are distinguished from
volved in the comprehension of non-poetic texts.
ample can be found in Reinhart (1976) which d
between two procedures involved in the comp
metaphor-focus interpretation vs. vehicle interp
suggests that whereas the former procedure may
both to the processing of poetic and non-poetic m
the second procedure that is involved in understa
metaphors.
Without considering the issue extensively, it is evident that
both approaches share the assumption that what distinguishes
poetic from non-poetic metaphor has nothing to do with the
internal semantic structure of these two types of figures, but
rather with their use. This common assumption suggests that,
in principle, the same metaphor can be defined as a poetic
metaphor in one context and as a non-poetic metaphor in an-
other; or to put it differently, that external considerations (that
is, external to the metaphor itself) determine whether or not it
is a poetic metaphor. Consequently, the implication is that the
poetic metaphor does not have a unique internal structure
which distinguishes it from non-poetic metaphor. Extrapolat-
ing from this position to other figures of speech which are
commonly held to be sample metaphors, such as the oxy-
moron, synesthesia, personification etc., it is commonly held
that the same attempt to distinguish between the poetic and
non-poetic figure obtains, namely, that the internal structure
of poetic figure is not different from that of the non-poetic fig-
ure.

Such a position is one that we would agree to acc

a last resort. The preferable theory (all other t
equal) is the one attempting to distinguish poetic f
etic figures in terms of an internal semantic stru
than in contextual terms. Only if we are convince
is no internal semantic structure of a given poetic figure
which can be distinguished from the non-poetic figure in
question, can we resort to the above solution.
In light of the above, the line of the argument to be devel-
oped in this paper can be outlined. Its central aim is to make a

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURE OF POETIC OXYMORON 107

first step towards the construction of a framework for handlin

the poetic/non-poetic distinction in terms of the internal se
mantic structure of the figure in question: the oxymoron. First
a distinction between two types of semantic structures will b
suggested. Both of these can, in principle, be exploited by an
oxymoron, i.e., the "direct" oxymoron vs. "indirect" oxy
moron. Second, these two structures will be compared with
respect to their frequency of use in a specific poetic corpus
This corpus consists of examples from Hebrew poetry as wel
as of several prototypical oxymora from English poetry. Com
paring the frequency of use of the above "direct vs. indirect
oxymoron in our poetic corpus, it will be argued that it
rather the "indirect" oxymoron's semantic structure that is
statistically dominant in the corpus, whereas the "direct
oxymoron's structure is statistically very infrequent. Since th
"indirect" oxymoron is the most frequent in our poetic corpus,
it may be characterized as the "poetic" oxymoron's structure
whereas the "direct" oxymoron will be characterized as the
"non-poetic" oxymoron. Subsequently, an accounting for th
difference in use will be suggested. The dominance of th
"poetic" oxymoron structure within the poetic corpus can b
accounted for by the more complex processing or understand
ing procedure(s) which this semantic structure requires com
pared to the other structure.
This methodology of distinguishing between the "poetic
and the "non-poetic" oxymoron should by no means be inter-
preted as implying that the non-poetic form characterizes th
use of oxymora outside poetic texts, that is, in non-poetic dis
course (although such a possibility is not excluded by the pre
sent argument). Rather, the label "non-poetic" simply mean
that its frequency in the poetic corpus is very low. However
the label "poetic oxymoron" means exactly what it says, tha
it is that semantic structure which is the dominant structure in
the poetic corpus in question. Obviously, this is a more mod
erate claim than that which correlates the distinction between
poetic and non-poetic structure with poetic and non-poetic lan
guage, and should, therefore, be considered as a first step to-
wards the definitive solution of the problem.

2. THE SEMANTIC STRUCTURE OF THE POETIC OXYMORON

2.1. Semantic Features
Provided that our aim is to draw a distinction between the
semantic structure of two types of oxymora (i.e., the "direct"
vs. "indirect"), the starting point should be a definition of the
object of research, the oxymoron.
Theories of poetic language usually define the oxymoron as

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
108 YESHAYAHU SHEN

a figure of speech consisting of two elements (o

which stand in "opposition," i.e., are antonymo
other (cf. Preminger 1975, Leech 1969, inter a
entirely different approach cf. Hrushovski 1984
relation of "opposition" is a semantic one betwee
ings of two lexical items, let us consider briefly
handled within a lexical semantic framework. One of the
well-known semantic theories which has been develo
the last 15-20 years, is Componential Analysis. Thi
postulates that meanings of lexical items which form
and theoretically infinite set, are, in principle, redu
relatively small set of "atoms of meanings" called sem
features or components (cf. Lyons 1977, inter alia). T
mantic features are conceptual units, the combinat
which can define the "meaning" of a given lexical item.
Thus, for example, the lexical item "man" is defined as a
combination of semantic features: "... +animate, +adult,
+male," whereas the lexical item "woman" is defined by t
same semantic features, except that the sign "+" is repla
by "-" for the last feature ("male").
The main characteristic of this analysis is that the sem
features are structured, that is, they are not randomly li
but rather are organized within an hierarchical structur
which some of the semantic features are higher than ot
This hierarchical structuring is significant, since the se
tic features of a given lexical item do not equally represen
meaning of that item. It is rather the lowest semanti
ture(s), which is(are) the "distinctive" one(s) and bears most
of the "semantic load" in that its function is to distinguish the
lexical item in question from its neighboring item. Thus, what
distinguishes "man" from "woman" is the lowest feature, i.e.,
the "+/- male" and not the other semantic features which are
identical.
Two basic semantic concepts which emerge from this th
ry are significant for the following discussion: the anton
and the hyponym. Two terms are antonyms when they s
all their semantic features save for a change in the "+/-" s
of their distinctive feature, e.g., "man"-"woman." A term
a hyponym of a given superordinate if its feature list incl
another one which is its distinctive feature, in addition to all
the semantic features of the superordinate term. Thus, the fea-

1. An entirely different approach to the definition of metaphor which can also be ap-
plied to other figures of speech is developed in Hrushovski (1984). Hrushovski's main
proposals rely on the assumption that notions like metaphor (and presumably the oxy-
moron, as well as other figures of speech) should be analyzed not as a linguistic units, but
rather as patterns which belong to the "textual semantics" level.

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURE OF POETIC OXYMORON 109

ture list of "bachelor" is derived from that of "man," n

"+animate, +adult, +male," to which the feature "-married" is
added. Such hierarchies are assumed to have some
psychological reality, as indicated by experim
(cf. Collins and Quillian 1972, Clark and Clark
and Johnson 1980).2
The meanings of antonym and hyponym bein
a distinction can be drawn between three semantic structures:
1. The "direct oxymoron" structure which consists of two
terms which are antonyms, namely, whose only differenc
consists of a change in the "+/-" sign of their lowest, distin
tive, feature, all others being identical.3 Examples of th
structure are "a feminine man," "living death" etc.
2. The "indirect oxymoron" structure in which one of its
two terms is not the direct antonym of the other, but rather t
hyponym of its antonym. Consider, for example, the phr
"the silence whistles" (taken from the Hebrew poet Nathan
Altherman's Summer Night) which is usually considered b
Israeli critics as a prototypical oxymoron in Hebrew poetry.
Its two terms are "silence" and "whistle." The feature list of
the first term,4 "silence," can be defined as (this is only
partial list): "+noun, +sensual, -count, ... -sound." The

2. Another point should be added regarding the psychological reality of the "featu
analysis." Various studies have raised arguments, supporting the "feature analysis" cla
for psychological validity. For example, an impressive correlation was found (cf. Malg
and Johnson 1980) between the number of features that two items share and the amou
of similarity which subjects tended to find between these items. This point also perta
to the present paper in that it substantiates the validity of the use of such notion
"availability" and of "cognitive distance" which are central to the semantic structure
oxymora.
3. Such a definition of the opposition relation can account for both the following com-
mon intuitions as to the meaning relations between two opposite terms: a) The intuition
that behind this opposition there is the largest possible similarity; this is explainable by
the fact that two opposites share all their semantic features, save one. b) The intuition that
despite the great similarity, the contrast between the opposed terms is the highest possi-
ble; this is accounted for by the fact that the "essence" of the "semantic load" is carried by
the lowest semantic feature(s).
4. The following point, regarding the issue of the "first and second terms" of a given
oxymoron, should be considered. "First/second term" are functional terms. The first
term of a given oxymoron is the starting point of the analysis, i.e., it is the first term
whose antonym is looked for. The method that has been used throughout the analysis
took the "comment" or "vehicle" of the oxymoron (usually the adjective) as the first term,
provided that it had a simple and straightforward antonym in the language, and the
"topic" or "tenor" (usually the noun) as the second term. In those cases where the
"comment"'s antonym was not lexically realized, or that there was no straightforward
path to it, it was the adjective (the second term), which was considered as the second
term. A case in point is the phrase "the silence whistles" in which the adjective
"whistles" does not have a straightforward antonym, and therefore the noun "silence"
was analyzed as the first term. However, in most of the samples analyzed in the paper,
the first term does have a straightforward lexicalized antonym.

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
110 YESHAYAHU SHEN

antonym of "silence" is lexically realized by the word

"sound" whose feature list consists of the same features for
"silence" save for the replacement of the "+" sign of th
tinctive feature "silence" (namely "-sound") by the "-" s
Note, however, that the second term of the oxymoron
"sound" but its hyponym, i.e., "whistle"; the feature list of
latter term adds the feature "+sharpness" to those of "so
and this addition turns "whistle" into a hyponym of "sil
Other examples from the Hebrew as well as English cor
are:

- "sacred garbage" (taken from the Israeli

Preil 1978). In this case the second term,
hyponym of the category "defile entitie
direct antonym of the first term "sacred."
-"cold fire" (Shakespeare, Romeo and Juliet). "Fire" is
the hyponym rather than a direct antonym of the
category "warm entities," which is the antonym of
"cold."
- "bright smoke" (Shakespeare, Romeo and Juliet). In
case the second term "smoke" is the hyponym of "d
which is the direct antonym of "bright."
- "sweet sorrow" (a typical oxymoron mentioned in Pr
minger 1975). Here the second term, "sorrow," is co
ceived of as an example (that is, a hyponym) of the
category "bitter entities"; the term "bitter" is the
antonym of the first term "sweet."
- "traitorous trueness" (Francis Thompson, The Hound of
Heaven). Here, the second term, "trueness," is not the
di'rect antonym of the first term antonym, "faithful," but
rather its hyponym.
3. The "metaphor" structure, since it is, roughly, common to
all metaphors. Here the two terms which comprise the phrase
do not differ in the sign "+/-" of the distinctive feature, or in
an additional feature, but in their "upper," that is, their "less"
distinctive features. Thus, one of the differences between
"silence" and "going" in the phrase "the silence goes" lies at
the upper level of the feature list: a higher level feature of the
term "going," "+movement" is not shared by the term
"silence."

2.2. The Semantic Structure of Poetic Oxymoron

In order to find which of these three semantic structures
characterizes the "poetic oxymoron," a random and large se
containing 100 samples of oxymora collected from the wri
ings of ten Israeli poets from the modern age of Hebrew po

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURE OF POETIC OXYMORON 111

ry was examined.5 They were judged by a number of nati

speakers as being samples of oxymora. Although this corp
seems relatively small to represent the "poetic phenomen
the following three points should be considered:
1. The samples were taken from poets who belong to tw
distinct periods of modern Hebrew poetry. Fifty samples w
composed by poets of the Revival Period while the others b
long to the Modern Period. The samples were, however, r
domly chosen, to avoid contextual restriction or biases, b
they of a particular text, poet, school of poets or of a given p
etics. It can therefore be assumed that these samples are i
dicative of extensive use of the poetic oxymoron.
2. To support the conclusions drawn from this corpus, 4
well-known samples collected from three literary dictionar
(Cuddon 1977, Shipley 1953, and Leech's Guide to English
etry [1969]) were examined. These samples are stated by t
authors to be the most typical samples of the oxymoron u
in poetry; moreover, as in Hebrew oxymora, these sample
were not restricted to a specific poet, poetics, or period. Th
characteristics reduce the risk of drawing too general conc
sions from a small set of data. It should be emphasized in
vance that the general tendencies revealed by the Hebrew
corpus are found also in the small corpus of English samples.6
3. The analysis presented here indicates the dominance of a
certain structure of the oxymoron in our corpus; it does not
impose a dichotomy between the poetic and non-poetic oxy-
moron, since the possibility of what might be characterized as
a "non-poetic" oxymoron appearing in a "poetic" text is by no
means precluded. Accordingly, the conclusions to be drawn
from this analysis should not be viewed as definitive or ex-
haustive, but rather as preliminary and initial indications in
support of a general direction of research still in progress.
This paper aims at drawing general parameters by means
of which the structure of the oxymoron should be described,
regardless of its specific context of appearance. A more de-
tailed study will have to examine how a specific context, e.g.,

5. The Israeli poets from whom the samples were taken are: H. Bialik, David Fogel, Ya-
cov Steinberg, Ya'acov Fichman and Ester Raab who belong to the Revival Period and
Nathan Altherman, Yocheved Bat-Miriam, Alexander Penn, Gabriel Preil and Leah
Goldberg who belong to the Modem Period.
6. Although there are certain differences between the Hebrew and English data, the
latter reveal the same tendencies as the former. These tendecies are even more impres-
sive considering the fact that the authors of the dictionaries from which these samples
were taken, define the oxymoron as consisting of antonyms. Thus, it is reasonable to as-
sume that their selection of typical oxymora was guided, a priori, by the tendency to look
for examples which generally confirm their definition.

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
112 YESHAYAHU SHEN

a given school of poetry, would determine the use of

rameters.

The main finding from the data was the foll

to what might be expected, only 16% of the
corpus were of the "direct" structure type, name
combines two antonyms. The most frequent st
"indirect," namely that in which the second
ponym of the first term's antonym, which ch
of the corpus.

3. A STRUCTURAL-COGNITIVE ACCOUNT FOR

THE SEMANTIC STRUCTURE OF THE "POETIC OXYMORON":
THE TWO CONSTRAINTS
3.1. The Problem
The problem is how to account for the fact that of the thr
possible semantic structures, it is the "indirect oxymor
that is prevalent in the poetic corpus.
I will propose a general explanation for this phenomen
which satisfies two (sub)questions: 1. Why is the "indire
oxymoron more frequent than the "direct" oxymoron?
Why is the "indirect" oxymoron more frequent than
metaphor?
It will be argued that these two questions can be answered
by the fact that the "indirect" oxymoron structure (rather than
the other two structures) meets two constraints: 1. It is con-
ceived of as an oxymoron (and not as a metaphor or as anoth-
er figure of speech). 2. Among the possible structures which
are conceived of as oxymora, it is the structure which requires
the most complicated processing.
The first constraint refers to the fact that the "indirect" oxy-
moron is an oxymoron, and it answers the second (sub)-
question by excluding the third structure (the metaphor) from
those possible in an oxymoron. Obviously, only the other two
structures (the "direct" and "indirect" oxymoron) meet the
first constraint.
The second constraint pertains to the fact that the "indirect"
oxymoron is poetic, which in this context means that it re-
quires complicated processing. The idea of equating
complexity of processing with "poeticality" is commonly
shared by theories of the poetic text, and its roots can be locat-
ed in the early works of the Russian formalists. Thus, the
"indirect" oxymoron rather than the "direct" oxymoron
meets the second constraint since, as will be demonstrated
below, it is the former that needs a more complicated
processing. Hence, this constraint answers the first
(sub)question formulated above.

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURE OF POETIC OXYMORON 113

To understand the notion "complexity of processing" c

tain cognitive considerations should be taken into account

3.2. Some Cognitive Considerations: the "Availability Scal

The association task is one of the main techniques used by
psychologists in order to obtain information concerning the
storage of lexical information in semantic memory. A subject
is presented with a stimulus word and is required to say the
first thing that comes to mind. His response is limited to a
single word (cf. Clark 1977). The relevance of these asso-
ciation tasks to the present paper is that they enable us to con-
struct the semantic relationship between lexical items on a
scale of "relatedness," based on their cognitive representation
in semantic memory. Accordingly, it can be argued that the
higher the "availability" of a given response, "b," in the con-
text of stimulus "a," namely, the probability of "b" being pro-
duced as a response to a stimulus word "a," the "smaller" the
"cognitive distance" between these two words in semantic
memory.7
According to Clark's paper, the principle that underlies a
large number of responses produced by subjects can be de-
fined as the "simplicity of production rule":
"Paradigmatic responses,7 therefore, appear to be produced
by a fairly homogeneous set of rules, perhaps ultimately by
one general rule. This simplicity of production rule might be
stated as follows: 'Perform the least change on the lowest fea-
ture, with the restriction that the result must correspond to an
English word.' Expanded, this rule defines 'least change' in
such a way that the operations of (1) changing the sign of a
feature, (2) deleting a feature, and (3) adding a feature, are of
increasing difficulty" (Clark 1970, pp. 280-281).8

7. Generally, Clark classifies the responses into two types, namely, paradigmatic vs. syn-
tagmatic responses. The former are those in which the output word maintains the syn-
tactic category of the stimulus (for example, "woman" which is a noun as a response to
"man," also a noun), whereas the latter involve a change in the syntactic category (for
example, the response "nice" to the stimulus "man"). In the present paper however, only
the principles underlying the paradigmatic responses are referred to because the
majority of the responses fell into this category.
8. The following quotation summarizes the general characteristics of the linguistic and
cognitive assumptions underlying the various semantic theories based on the idea of
"decomposition," i.e., the idea that a meaning of a lexical item can be decomposed into
semantic primitives: "In general ... Literalist approaches (i.e. those based on the decom-
position assumption-Yeshayahu Shen) involve the following core claims (whose precise
nature varies with the particular model): Primitive elements (e.g., features, concepts,
propositions) are said to exist in memory-the elements postulate. Words are repre-
sentable in memory as a static collection (i.e., a dictionary) of elements-the dictionary
postulate. The elements are related in terms of links or paths bearing labels describing
the nature of the relationship (e.g., case relation, part-whole) and varying in their

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
114 YESHAYAHU SHEN

The relationships between the stimulus and r

association enable us to distinguish between
each of which represents an increasingly com
cessing procedure. The least complicated in the
scale" is the changing in the sign of the featu
the deletion of a feature, and the most compli
the addition of a feature. Marshall's theory (M
can explain this scale, since it shows that it is e
a semantic feature than to add one because the
candidate for the deletion, i.e., the distinctive feature, whereas
there are several candidates for addition. Thus, the construc-
tion of "animal" from "dog" requires a deletion of only that
feature which distinguishes dogs from other animals, whereas
the construction from "animal" to "dog" is more complicated
because there are several potential responses (dog, cat, cow,
horse etc.).
These procedures are characterized by being based on one
"processing move" (a change in the sign, a deletion or an
addition of a semantic feature). Analyzing the data reported in
Clark and Clark (1977) it is possible to add another option to
these three: the most complicated one, namely, that one based
on two processing moves (such an option is not described by
the authors). In such a case, in order to construct the response
out of its stimulus, the speaker changes the stimulus term from
"+" to "-," or vice versa, yielding a list of semantic features to
which he makes an addition or a deletion. For instance, if one
moves from the stimulus "man" to the response "girl," one
uses the above option: first, the sign of the lowest feature of
"man," i.e., "+male" is changed into "-male" producing the
distinctive feature of "woman"; then an additional lowest fea-
ture is produced: "-adult," which is the distinctive feature of
"girl." As the data clearly indicate, the least frequently occur-
ring option is the fourth one. Thus, in the case of the stimulus
"man," the response "girl" occurs only in 3% of the total sum,
whereas "boy" occurs in 8% and "woman" in 62% (this order

directionality-the link postulate. Words that are semantically similar are "closer to-
gether" in memory than are disjoint words, that is, distance is a direct junction of ele-
ments overlap-the distance postulate. The labels or descriptions on the paths place re-
strictions upon possible element combinations-the restriction postulate. Elements com-
bine in a compositional, non-Gestalt manner-the compositionality postulate. Remem-
bering constitutes an attempt to match input elements or element structure with those al-
ready stored; stored elements are usually content-addressable, and matching is a matter
of compatibility of input element structure with memory element structure-the match-
ing postulate. Outputs (recall, true-false judgments, etc.) reflect knowledge as a verification
process- the verification postulate (the ultimate form of the verification view is procedural
semantics which replaces the proposition as the basic element or sense of a linguistic
unit, with mental procedures for deciding when the unit applies to an event ...)" (p. 129).

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURE OF POETIC OXYMORON 115

is maintained in other examples used in the association t

reported by Clark [1977]).
Having obtained the above results we may establish
following "availability scale" of a given lexical item in
context of another: the most available is the antonym, the s
ond degree of availability is assigned to the superordinate
third to hyponym, and the fourth to the antonym plus sup
dinate (or hyponym).

3.3. Processing Complexity

Let us now turn from these cognitive aspects of process
lexical items to our initial purpose, the construction of a
ry that will explain how an oxymoron, that is, a phrase w
combines two lexical items, is processed. In order to inco
rate the above "availability" scale into a theory of the
processing of figures of speech, an additional assumption is
required, according to which such processing is based on a
"cognitive search" wherein the processor attempts to locate in
his "semantic memory" the semantic features shared by both.
Having in mind the "availability" scale we may assume that
the complexity of processing a phrase which consists of two
terms, depends on the "availability" of one of the terms in the
context of the other.
The three semantic structures previously discussed can be
ranked with respect to their processing complexity:
1. The least complicated processing is required by the
"direct" oxymoron, namely, that structure which consists of
two antonyms. The reason for this is that in the context of a
given term, the most available, and therefore the term which
requires the least effort of "cognitive search," is its antonym.
2. The next degree of "complexity" is required by the
"indirect" oxymoron, in which the second term is the hy-
ponym of the first term's antonym. This hyponym is of a low-
er availability in the context of the first term, since it is the
antonym itself which is of the highest availability.
3. The most complicated processing is required by the
"metaphor" structure, where the second term is the least
available in the context of the first one. A case in point is the
phrase "a dog-like man" which presents a "metaphor" struc-
ture (the difference between simile and metaphor is of no rel-
evance to this case), in which the meaning of the second term
(i.e., "dog," in the phrase "dog-like") differs from the mean-
ing of the first term not only in its lowest, i.e., distinctive, fea-
ture, but also in the higher ones on its hierarchical list, such
as "+/-human."
We are now in a position to return to the initial constraints

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
116 YESHAYAHU SHEN

we imposed. Because it is not an oxymoron structure, the

"metaphor" structure is excluded by the first constraint. The
complementary (second) constraint required that from among
the other two structures the "poetic" oxymoron should have
the most complicated structure to process. Clearly, among the
first two structures it is the "indirect" oxymoron which is
more complicated to process. In sum, the "indirect" oxy-
moron is the only structure which meets both constraints, be-
ing the most complicated structure to process which can still
be considered as an oxymoron.

4. THE PROTOTYPE: ANOTHER SCALE OF PROCESSING COMPLEXITY

4.0. An Introduction
Having outlined the characteristics of the poetic oxymoron
terms of the cognitive and structural constraints which de
mine its specific semantic structure, we still face the follo
ing. It will be recalled that it was postulated that the sema
structure of the poetic oxymoron is characterized as one i
which one of the opposed terms is a hyponym of the anton
of the other term. A given superordinate category, howe
usually dominates more than one hyponym (or subordina
term); thus, for example, the superordinate category "sou
has several subordinate members which are specific types
sounds, such as "cry," "whistle," "shouting" etc. In other
words, the semantic structure of the poetic oxymoron leaves
the producer of a given oxymoron a choice among a range of
alternatives as to the actual hyponym in the position of the
second term.
In order to complete the description of the semantic s
ture of the poetic oxymoron, we shall have to consider
second aspect of its semantic structure. Since such a de
tion involves a choice between hyponyms, the crucial q
tion is whether this choice is regulated by systematic te
cies. In other words, can some constraint be identified and
imposed on the semantic structure of the poetic oxymoron
thus specifying its form? In the following sections this issue
addressed.

4.1. The Poetic Oxymoron and the Notion of the Prototype

The structural regularities to be described, require som
nitive considerations of the notion of the "prototype" t
from Rosch's theory (1978).
Against the traditional approaches to categorical orga
tion in memory (cf. Collins and Quillian 1972), Rosch a
that the members dominated by a given superordinate c
ry are not equally stored in memory: some members a

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURE OF POETIC OXYMORON 117

"prototype," that is, they are, relative to other members, "g

examplars" of the whole set of category members. Consi
for example, the superordinate category "furniture" and
members that comprise this category, e.g. "chair," "tabl
"footstool," "rug" and "curtain." These members cannot
considered as equal representatives or examplars of the s
perordinate category: whereas table and chair are consider
to be "good," the footstool is a relatively "medium" exam
and rug and curtain are "very poor" examplars of the supe
dinate category. These "very poor" examplars are conside
as standing on the fuzzy border line which distinguishes
category in question from its neighbors, and sometimes t
are not even considered as included within the categor
question (cf. Cohen and Murphy 1984, a paper discussing
pects of ranking degrees of typicality or prototypicality of
egory members according to various scales).9 Another ex
ple is the category "bird" of which "chicken" is a rather
example; it is argued (see Rosch and Mervis 1975) that so
speakers will hardly identify this poor example as a mem
of the category "bird." (The importance of this fact will
specified in the subsequent discussion.)
The characteristic of the prototype most relevant to
discussion, is its high "availability" in the context of its su
ordinate category (see Rosch 1978); (henceforth the term
"prototype" will refer to the "good" examplar of a given cate-
gory). Thus, subjects who were given a category name and
then asked to provide examples of it, tended to respond with
the prototypes rather than the poorer examples, which indi-
cates the higher availability of prototypes in comparison with
other members of the category. It may be argued that the
"cognitive distance" between a given term and its superordi-
nate category depends on the prototypicality of the former in
the context of the latter: the cognitive distance (and hence the
processing complexity) decreases as the degree of prototypi-
cality increases.
Returning to our initial consideration regarding the seman-
tic structure of the poetic oxymoron, the foregoing description
leads to the construction of another hierarchy in which vari-
ous (sub)types of "poetic" oxymora can be distinguished ac-
cording to their processing complexity. We have defined the
"poetic" oxymoron as consisting of two terms, the second of

9. Mervis and Rosch (1981) characterize the "goodness" of a given examplar relative to
the amount of features that it shares with the other members of the set. Thus, the
"prototype" is that member in a given set that shares the maximal number of features
with the other members in that set, whereas the "poor" examplar shares a relatively
small amount of these features.

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
118 YESHAYAHU SHEN

which is the hyponym of the first one's antonym

notion of prototypicality in mind, it may be assum
ponyms differ in their typicality with respect t
perordinate category.
Three general types of "poetic" (or "indirect
can be distinguished according to processing comp
1. The unmarked structure, which is typical of t
oxymora in which the hyponym of the antonym
term is a prototypical example of the superordina
The poetic oxymoron "the silence cries" illustrate
since the hyponym "cries," the second term, is a
example of its superordinate category "sound," the
antonym. A further example is "cold fire" in w
the second term, is a prototypical example of t
"hot" which is the first term's antonym. Due to
availability of its hyponym, this structure require
degree of processing complexity in comparison
types below.
2. The medium structure requires more complex processing
where the hyponym is a "medium" example of the superordi-
nate category, namely, all the examples that are between the
prototype pole, on the one hand, and the "very bad examplar"
pole, on the other. A case in point is the oxymoron "the si-
lence whistles" in which the hyponym "whistles" is con-
ceived of as a "medium" example of the superordinate
category "sound." (The intuitive criteria of ranking a given
hyponym as a "medium" example is the question whether
better and poorer examples for the category "sound" can be
found. Thus, in the case of "whistle" it is reasonable to assume
that some "sounds" are better examples of the category
"sound," e.g. "cry" and "shout" while others are poorer, e.g.,
"sigh"; according to the above intuitive criteria, we may clas-
sify "whistle" as a "medium" example of the category
"sound." Other examples from our corpus are:
- "sacred garbage." In this case the antonym of the adjec-
tive "sacred" is something like "defiled" or "impure"
(in the religious sense); the second term "garbage" is a
"medium" example of "defiled."
-"bright smoke." The noun "smoke" represents a
"medium" example of the category "dim" which is the
antonym of "bright."
_ "sweet sorrow." The noun "sorrow" represents a
"medium" example of the category "bitterness."
"traitorous trueness." The noun "trueness" represents a
"medium" example of "faithfulness."

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURE OF POETIC OXYMORON 119

3. The marked structure, the most complex structure to pr

cess, is one in which the hyponym is a "very bad examp
of its superordinate category, namely, that there is no p
example to its superordinate category, or that it is at least v
difficult to find one. Thus, in the phrase "the silence si
the hyponym "sighs" is a very bad example of the catego
"sound." Another example taken from our corpus is the
moron "the shining (or glamorous) suffering" (taken fr
Bat-Miriam 1972). In this case the first term is "shining
"glamorous" which is supposed to be the hyponym of
"delight" or "pleasure," the antonym of "suffering." Note,
however, that "glamorous" is a very poor example (or even a
non-member) of the category "delight." (All the typicality
judgments were provided by native Hebrew speakers who
were asked to rank these samples.)
On the basis of this scale, those oxymora which were de-
fined as consisting of an "indirect" oxymoron structure were
examined. The main finding was that the "medium case" type
structure is prevalent in the corpus examined, whereas a rela-
tively small amount of the other two possible structures, i.e.
the "marked" and the "unmarked," were found. Out of a total
sum of 85 "indirect" oxymora, 57 (67%) were of the "medium
case" type, 19 (22%) were of a structure which can be ranked
as unmarked, and 9 (11%) were of a typical "marked" struc-
ture.

4.2. An Account for the Hyponym-superordinate Relations

the Poetic Oxymoron
These findings relative to hyponyn-superordinate relati
can be accounted for by the fact that the "medium" structu
the only one which meets the two constraints described earl
er: 1. It is conceived of as an oxymoron and not as a metap
2. Among the possible structures which are conceived of
oxymora it is the one which requires the most complica
processing.
Note that the first constraint is met only by the first
structures but not by the third, "marked" structure. The re
is that in the case of the "marked" structure, the hypon
stands on the fuzzy border line of the superordinate categ
yielding a structure which in turn stands between oxym
and metaphors, and in any case it cannot be conceived of
equivocally as an oxymoron. Support for this claim was f
by asking five native speakers of Hebrew for two indepen
judgments: 1. Whether a given phrase, e.g., "the silen
sighs" or "the glamorous suffering," is an oxymoron

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
120 YESHAYAHU SHEN

metaphor, and 2. Whether the hyponym, e.

"glamorous" respectively, can be considered a m
superordinate category i.e., "sound" and "deligh
The results indicated a strong correlation betw
identification of a given phrase as an oxymoro
judgment as to whether the hyponym in quest
included in the prototypicality scale. Thus, it was
The three speakers who judged the hyponym i
something which is "between a very poor exam
member of the category" or that is totally excl
ble member of the above category, were those
the definition of the entire phrase as an oxymo
it to be a metaphor; and 2. The two speakers w
poor member within the above category, defin
(glamorous) suffering" as "something between
and non-oxymoron"; they did not see it as an o
se. (Typically, it took them more reaction time
the above question than in other cases in whic
asked to determine whether a given phrase was an oxy-
moron.) In addition to the data presented in Rosch's work,
these provisional and initial findings provide some support to
the claim that the "bad examplar" stands on the border line
between the given category and another.
These findings support our claim that the "marked" struc-
ture stands on the border line between oxymoron and
metaphor and that it does not therefore meet the first con-
straint which requires it to be unequivocally seen as an oxy-
moron.

As for the second constraint, namely, the

the oxymoron in question be cognitively th
structure, it is evident that among the firs
tures, i.e., those which meet the first constr
"medium" structure which meets it, becaus
ly explained, it requires the more complica
Our main finding, namely, that within
oxymora in our corpus, it is the "medium
the highest frequency of occurrence, is, th
for by our two constraints: 1. The "medium
the most complicated processing among th
2. It can be unreservedly conceived of a
moron.

SUMMARY AND CONCLUSION

The general picture outlined in this paper suggests that the s
mantic structure of the poetic oxymoron is a structure whic
can be characterized on two levels of analysis: 1. On the fir

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
 STRUCTURE OF POETIC OXYMORON 121

(the relations between the two explicitly stated terms of

oxymoron), it is based on "indirect antonymous" relation
between the two terms, i.e., a structure in which the second
term is the hyponym of the antonym of the first term (and not
its direct antonym). 2. On the second level (the relations be-
tween the hyponym and the constructed antonym) it is char-
acterized by the fact that the hyponym is the "medium exam-
ple" of the constructed antonym. In order to account for these
structural characteristics, I have suggested, for both levels of
analysis, that there are two constraints which are imposed on
the semantic structure of the poetic oxymoron.
The two constraints, the reasons for their existence, as well
as the way in which they are integrated into the framework of
literary theory, should by now be evident. The combination of
both constraints implies that the poetic oxymoron is that
which requires the most complicated processing possible
within the limits of that figure of speech, i.e. that the phrase in
question can still be counted as oxymoron. The reason for
these constraints is based on the idea, well known in literary
theory, that the poetic phenomenon (in our case the poetic fig-
ure of speech) requires a more complex processing or under-
standing procedure than non-poetic linguistic phenomena.
Thus, among those structures that can be counted as oxymoron
structures, the most prevalent is that which requires the most
complicated processing.
The starting point of this paper was an attempt to distinguish
between poetic and non-poetic oxymora in terms of their in-
ternal semantic structure. It may be stated that this attempt has
led us not only to the conclusion that a characterization of the
semantic structure of the poetic oxymoron is indeed possible,
but also that the high frequency of this structure can be moti-
vated by the very attempt of the poetic utterance to complicate
its comprehension and processing. This paper should be
viewed as a manifestation, based on advances in semantic and
psycholinguistic theory, of the well-known tenet of literary
theory that views the poetic utterance as one which aims to
impose various complications and difficulties on the reader
throughout the comprehension process.
REFERENCES
Beardseley, M.C., 1958. Aesthetics (New York).
Bat-Miriam, Y., 1972. Between Sand and Sun (Tel Aviv: Hakibbutz Hameuhad).
Black, Max, 1962. "Metaphor," Models and Metaphors. (Ithaca: Cornell UP).
Clark, H.H. and E.V. Clark, 1977. Psychology and Language (New York: Harcourt
Jovanovich, Inc.).
Clark, H.H., 1970. "Word Association and Linguistic Theory," in: J. Lyons, ed.,
Horizons in Linguistics (Harmondsworth).
Cohen, B. and L.G. Murphy, 1984. "Models of Concepts," Cognitive Science 8, 27-58.

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms
122 YESHAYAHU SHEN

Collins, A.M and R.M. Quillian, 1972. "Experiments in Semanti

guage Comprehension," in: W. Lee Gregg, ed., Cognition
Memory (New York, London, Sydney, Toronto: John Wiley a
138.

Honeck, R.P., 1980. "Proverbs, Meaning, and Group Structure," in: R

Hoffman, eds., Cognition and Figurative Language (Hillsdale,
161.

Hrushovski, Benjamin, 1984. "Poetic Metaphor and Frames of Ref

day 5:1.
Lyon, John, 1977. Semantics 1 and 2 (Cambridge UP).
Malgadi and Johnson, 1980. "Measurement of Figurative Language: Semantic Feature
Models of Comprehension and Appreciation," in: R.P. Honeck and R.R. Hoff-
man, eds., Cognition and Figurative Language (Hillsdale, NJ: Erlbaum).
Marshall, J., 1969. "Psychological Linguistics: Psychological Aspects of Semantic Struc-
ture," in: A.R. Meetham, ed., Encyclopedia of Linguistics, Information and Con-
trol (London: Pergamon Press), 442-444.
Mervis, C.B. and E. Rosch, 1981. "Categorization of Natural Objects," Annual Review of
Psychology 32:89-120.
Preil, G., 1978. Poems (Tel Aviv: "Yachdav").
Reinhart, T., 1976. "On Understanding Poetic Metaphor," Poetics 5, 383-401.
Rosch, E., 1975. "Cognitive Representations of Semantic Categories," Journal of Experi-
mental Psychology: General, 104, 192-233.
1978 "Principles of Categorization," in: E. Rosch and B. Lloyd, eds., Cognition and
Categorization (Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers).
Rosch, E. and C.B. Mervis, 1975. "Family Resemblances: Studies in the Internal Struc-
ture of Categories," Cognitive Psychology, 7, 573-605.
Tzur, R., 1983. "The Cognitive Theory of Metaphor," Mechearey Yerushalayim (in He-
brew).

This content downloaded from

193.136.113.72 on Sat, 26 Mar 2022 00:48:05 UTC
All use subject to https://about.jstor.org/terms