Conceptual metaphor theory

Some criticisms and alternative proposals

Zoltn Kvecses
Etvs Lornd University, Budapest

Despite its popularity in and outside cognitive linguistics, cognitive metaphor

theory (CMT) has received a wide range of criticisms in the past two decades.
Several metaphor researchers have criticized the methodology with which
metaphor is studied (emphasizing concepts instead of words), the direction of
analysis (emphasizing a top-down instead of a bottom-up approach), the cat-
egory level of metaphor (claiming its superordinate status instead of basic level),
the embodiment of metaphor (emphasizing the universal, mechanical, and
monolithic aspects instead of nonuniversal, nonmechanical, and nonmonolithic
aspects of embodiment), and its relationship to culture (emphasizing the role
of universal bodily experience instead of the interaction of body and context).
In the paper, I respond to this criticism largely based on my own research and
propose a view on these issues that can successfully meet these challenges and
that can be regarded as an alternative to the standard theory.

Keywords: metaphor, embodiment, culture, words vs. concepts, bottom-up vs.

top-down approach, basic level vs. superordinate level

The theory of metaphor proposed by George Lakoff & Mark Johnson (1980),
henceforth conceptual metaphor theory (CMT), has been criticized for several
reasons and from several perspectives in the past 25 years. Of these, I will select
and briefly deal with five large issues in the present paper. These are: (1) the issue
of methodology; (2) the issue of the direction of analysis; (3) the issue of schema-
ticity; (4) the issue of embodiment; and (5) the issue of the relationship between
metaphor and culture.

I. The issue of methodology

One of the most often heard criticisms of CMT is that most researchers in CMT
set up conceptual metaphors on the basis of intuitive and unsystematically found

Annual Review of Cognitive Linguistics 6 (2008), 168184. doi 10.1075/arcl.6.08kov

issn / e-issn  John Benjamins Publishing Company
 Conceptual metaphor theory 169

linguistic metaphors (Pragglejaz Group, 2007). That is to say, the charge is that
many CMT researchers examine their own mental lexicons or the data found in
dictionaries and thesauri, and on the basis of some linguistic examples they ar-
rive at and suggest conceptual metaphors. If, for example, in the dictionary the
verb boil also means to be very angry, explode means to lose control over anger,
hotheaded means someone who loses control over anger easily, seething means to
lose control over anger at any moment, then CMT researchers conclude that there
exists a conceptual metaphor that we can call ANGER IS A HOT FLUID IN A
CONTAINER (Lakoff & Kvecses, 1987).
Those who criticize this approach find this procedure problematic in two ways.
On the one hand, they claim that CMT researchers take for granted which expres-
sions are metaphorical, and, on the other, critics also suggest that the approach
does not pay attention to which actual expressions are used for the target domain
(of, for example, anger) by real speakers in natural discourse (see, for example,
Pragglejaz Group, 2007). They propose, in light of these objections, that we need
to construct a reliable methodology to identify metaphorical expressions and we
need to use real corpora in the course of identifying such expressions.
Though I am largely in agreement with both objections, I do not think that ei-
ther of these criticisms invalidates the practice of CMT as described above. In my
view, the systematic identification of metaphors in natural discourse and corpora
does not occur at the same level of metaphor analysis where a large number of
cognitive linguists (including myself) work.
In a cognitive linguistic approach, three levels of the existence of metaphors
can be distinguished: the supraindividual, the individual, and the subindividual
levels (see Kvecses, 2002, chapt. 17). At the supraindividual level, we find decon-
textualized metaphorical linguistic expressions (e.g., in dictionaries) on the basis
of which we can suggest certain conceptual metaphors. At the individual level,
specific speakers use specific metaphorical linguistic expressions in specific com-
municative situations in relation to particular target concepts. The subindividual
level is the one where the metaphors receive their motivation, that is, the meta-
phors have a bodily and/or cultural basis.
In other words, the systematic identification of linguistic metaphors in natural
discourse is a goal that is connected with what I call the individual level. For this
reason, the objections do not invalidate the goals of the supraindividual level: to
propose conceptual metaphors on the basis of linguistic expressions that research-
ers intuitively take to be metaphorical. The two levels are associated with different
goals. At the same time, however, the goals of the two levels complement each
other, in that the metaphors suggested on an intuitive basis may prove to be useful
in organizing the systematically identified linguistic metaphors into larger con-
ceptual metaphors used at the individual level and, also, because the systematically
170 Zoltn Kvecses

identified linguistic metaphors in real discourse may lead to the discovery of so far
unidentified conceptual metaphors.

II. The issue of the direction of analysis

Closely related to the previous criticism, several critics raise the issue of the di-
rection of analysis, that is, the issue whether the analysis of metaphors should
be done top-down or bottom-up (Dobrovolskij & Piirainen, 2005; Stefanowitch,
2007). Clearly, those researchers who follow the traditional practice of CMT
type of analysis follow the top-down direction, since on the basis of a small num-
ber of decontextualized examples they postulate conceptual metaphors and then
they examine the internal structure of these metaphors (set up mappings, entail-
ments, etc.). In such an approach, what is in the center of attention is the concep-
tual metaphor itself as a (hypothetical) higher-level cognitive structure. By con-
trast, in bottom-up approaches a large number of expressions are studied (e.g., an
entire corpus), the metaphorical expression are identified on the basis of a well
established protocol (Pragglejaz Group, 2007), the metaphorical expressions are
checked for their detailed behavior (semantic, structural, pragmatic, esthetic, etc.)
in concrete contexts of use, and finally conceptual metaphors are established as a
result of a multi-stage procedure (see, e.g., Steen, 1999). In this kind of approach,
what is in the center of attention is language and linguistic metaphors, as well as
their behavior in specific contexts. The representatives of the latter approach usu-
ally raise two objections against the top-down approach. The first is what can be
called the principle of the dominance of irregularity, and the second is the goal to
identify each and every linguistic and conceptual metaphor relating to a particular
target domain.
If we are primarily concerned with linguistic structures and processes (as op-
posed to hypothetical cognitive structures and processes), then in the corpora we
are working with we are likely to find irregularities, rather than regularities. The
principle of the dominance of irregularities means that the individual metaphori-
cal expressions will be found predominantly irregular as regards their semantic
behavior despite the fact that, in the main, they come into existence as a result of
regular cognitive processes, such as conceptual metaphors (Dobrovolskij & Pi-
irainen, 2005). That is to say, the main objection is that the top-down approach
that emphasizes regularities (like conceptual metaphors) cannot account for the
unique and irregular semantic behavior of many metaphorical expressions. Do-
brovolskij & Piirainen state:
 Conceptual metaphor theory 171

Central to our approach is the idea that even if the production of CFUs [conven-
tional figurative units] is governed by some general principles of human cogni-
tion, they remain, above all, irregular units of the lexicon. Thus, the most salient
features of their semantic structure and discursive behaviour cannot be captured
by metalinguistic tools aimed at exclusively discovering regular characteristics.
Large portions of CFUs came into being under the influence of certain culture-
specific phenomena. (Dobrovolskij & Piirainen, 2005, pp. 355356)

As can be seen, according to Dobrovolskij & Piirainen, the reason for the irregu-
larity is that the assumed irregular metaphorical expressions (such as metaphor-
based idioms) have often emerged from and have been shaped by certain culture-
specific factors and influences, and therefore, they cannot be explained by larger,
often universal conceptual metaphors.
In other words, a major problem that representatives of the bottom-up ap-
proach point out in connection with the top-down approach is that conceptual
metaphors do not provide an account of the meaning and, more generally, the
irregular linguistic-semantic behavior of many metaphorical expressions. These
critics of traditional CMT believe that the linguistic behavior of metaphors is more
irregular than regular, and that this dominant feature of linguistic metaphors is
hidden by an approach that emphasizes (hypothetical) global cognitive structures,
such as conceptual metaphors.
The criticism is partly valid, partly not. Let us take as an example the meta-
phorical expression split hairs. How can we explain the meaning of the expression
(i.e., its semantic behavior) in light of traditional CMT? In order to be able to
do that, we would have to know which global conceptual metaphor underlies the
expression; that is, the metaphor that motivates the meaning to pay too much
attention to small and unimportant differences in an argument. It is not easy to
find a global conceptual metaphor that can readily and naturally account for this
meaning. This situation is unlike cases where we have well known metaphors, such
as ANGER IS A HOT FLUID IN A CONTAINER (for the word boil), THEORIES
ARE BUILDINGS (for foundation), or LIFE IS A JOURNEY (for bumpy road),
that easily come to mind and readily explain the behavior of many metaphorical
expressions. If there is no easily retrievable conceptual metaphor that accounts for
the meaning of the expression split hairs in a natural way, then the expression will
remain hopelessly unmotivated, that is, irregular, as regards its meaning. But even
if there is such a global metaphor in this case (though its not easy to find), it is
likely that there are many other cases where no global conceptual metaphors can
be discovered and used to explain the semantic behavior of particular metaphori-
cal words and expressions. Given this, we cannot, at least in my judgment, claim
that there is a global conceptual metaphor behind, or underlying, each and every
metaphorical expression. Thus, in this respect, I find the criticism valid.
172 Zoltn Kvecses

But, in another respect, it is not valid. In Dobrovolskij & Piirainens (2005)

view, CMT cannot in many cases account for the dissimilar behavior of meta-
phorical linguistic expressions that are instances of the same conceptual meta-
phor. According to Dobrovolskij & Piirainen, the fact that two or more linguistic
expressions belong to the same conceptual metaphor does not explain the dif-
ferent meanings these expressions have. Let us take the following expressions as
examples: add fuel to the fire and flare up, as in His stupid comment just added
fuel to the fire and The argument flared up between them. Although we can be
sure that what motivates the expressions is the ARGUMENT IS FIRE conceptual
metaphor, Dobrovolskij & Piirainen would maintain that the difference in their
meaning is not explained by the conceptual metaphor alone. The meaning to in-
crease the intensity of the argument (in the case of add fuel to the fire) and the
meaning the intensity of the argument increases suddenly (in the case of flare
up) are significantly different, but the difference is not captured by the (theory of)
conceptual metaphor.
What Dobrovolskij & Piirainen leave out of their account is that CMT is not
exhausted by setting up global conceptual metaphors, like ARGUMENT IS FIRE,
or ANGER IS A HOT FLUID IN A CONTAINER, or THEORIES ARE BUILD-
INGS. As a matter of fact, the setting up of such metaphors is merely a first step in
the course of the analysis. What we need to do in addition is to see which elements
of the source correspond to which elements of the target domain. These are the
correspondences (or mappings) that crucially constitute conceptual metaphors.
In the case of the ARGUMENT IS FIRE metaphor, the mappings display a com-
plex fine-grained structure. Since the ARGUMENT IS FIRE metaphor is part of a
more general metaphor (INTENSITY IS HEAT), it is appropriate to deal with the
mappings of the ARGUMENT IS FIRE metaphor within the framework of this
more general, global metaphor (see Kvecses, 2002, 2005). In a simplified form
and leaving out many details, the INTENSITY IS HEAT metaphor consists of the
following mappings:
Source: Target:
the degree of heat the degree of intensity
the cause of heat the cause of intensity
increase in the degree of heat increase in the degree of intensity
decrease in the degree of heat decrease in the degree of intensity
heat drops to zero intensity ceases
Of these mappings, different ones will apply to His stupid comment just added
oil to the fire and The argument flared up between them. The meaning of the
expression add fuel to the fire is based on and motivated by the mapping the cause
 Conceptual metaphor theory 173

of heat the cause of intensity, whereas the meaning of the expression flare up is
based on and motivated by the mapping increase in the degree of heat increase
in the degree of intensity. Although this does not fully account for the meaning of
the expressions, we can see how the detailed and specific mappings can systemati-
cally motivate subtle differences between the semantics of linguistic expressions
that belong to the same conceptual metaphor.
Let us now turn to the second objection raised against the top-down approach.
This is the suggestion that our overall goal in metaphor analysis should be to find
each and every linguistic and conceptual metaphor relating to a target in a given
corpus. The claim is especially characteristic of the practitioners of corpus linguis-
tics (see, for example, Stefanowitch, 2007). The deeper question that we need to
ask in this connection is why, indeed, it is so important for us to find metaphors for
particular target domains. Is it because it is inherently good to find the complete
repertoire of metaphors and to produce a complete list of them or is it because
such list could be used for some useful purpose? No doubt, to have such lists could
serve very useful purposes, but I believe that setting up a complete list is not the
most important reason for studying metaphor in cognitive linguistics. The most
important reason, I propose, is to be able to see to what extent and with which
content the metaphors contribute to the conceptualization of abstract concepts, as
well as their cognitive representation. After all, our ultimate goal is to understand,
as best as we can, the mind that consists, in part, of a large number of concepts that
function as target domains for an equally large number of source domains.
The corpus linguistic approach can identify a great deal more metaphorical
expressions for particular target domains than the intuitive, traditional approach
can (e.g., Stefanowitch, 2007). At the same time, however, this quantitative ad-
vantage does not necessarily lead to a qualitative advantage. Stefanowitch (2007)
examined five words (anger, fear, happiness, sadness, and disgust) related to emo-
tions in the British National Corpus. His results, in the main, support the results
of top-down researchers: the most characteristic metaphor for anger is heat as a
source, the one for happiness it is an upward orientation, and so forth.
The corpus-linguistic approach did not find most of the specific-level emotion
metaphors as being significantly unique to particular emotion concepts. At the
same time, these are the metaphors that serve as the basis of our knowledge about
emotion concepts. Metaphors such as EMOTIONS ARE OPPONENTS, EMO-
TIONS ARE NATURAL FORCES, EMOTION IS INSANITY, and so forth, define
key aspects of emotion concepts (such as control, passivity, lack of control) (see
Kvecses, 2000/2003). This discrepancy between the results makes it clear that our
intuitions (in this case, about emotion concepts) are inevitable for the interpreta-
tion of the results produced by corpus linguistics.
174 Zoltn Kvecses

Generic-level metaphors make up the majority of emotion metaphors in Ste-

fanowitchs study. These are such Event Structure metaphors as (EMOTIONAL)
STATES ARE CONTAINERS, (EMOTIONAL) STATES ARE OBJECTS, CAUSES
(OF EMOTIONS) ARE FORCES, and (EMOTIONAL) CHANGE IS MOTION.
These generic-level metaphors apply to any state, cause, or change, and for this
reason they have limited usefulness in characterizing the specific content and
structure of prototypical models of emotion concepts. This is so even though we
find several interesting differences between emotion concepts at the level generic
metaphors.
In sum, as regards the first objection concerning the direction of analysis, I
suggest that there are far fewer irregularities in linguistic metaphors than what
is proposed by the critics if we take into account the large systems of mappings
that characterize individual conceptual metaphors and their hierarchical systems
(Kvecses, 1995, 2000). With the help of detailed systems of mappings we can ac-
count for a large number of subtle meaning differences. According to the second
objection, the identification of linguistic and conceptual metaphors should be as
complete as possible. But, as we have seen above, quantity does not guarantee that
we can achieve our most important goal better; quantitative metaphor analysis
must be supplemented by intuitive qualitative analysis. (A more detailed response
to Stefanowitchs critique can be found in Kvecses, forthcoming.)

III. The issue of schematicity

Several critics raise the issue of the schematicity of metaphors, that is, at which
level of schematicity we should formulate conceptual metaphors. For instance,
Clausner & Croft (1997) note that the well known conceptual metaphor intro-
duced by Lakoff and Johnson (1980) THEORIES/ ARGUMENTS ARE BUILD-
INGS does not generalize the linguistic facts at the appropriate level. We can say,
for example, that The theory has a solid foundation, but we cannot say that The
theory has long corridors and high windows. This means that it is not the case that
each element of the source can be utilized to talk about the target. For this rea-
son, Clausner & Croft (1997) propose a less schematic version of this metaphor:
THE CONVINCINGNESS OF THEORIES/ ARGUMENTS IS THE PHYSICAL
INTEGRITY OF THE BUILDING. In their words:
However, the metaphor is actually less schematic in both source and target do-
mains than this new formulation (rational arguments are buildings). the
formula the convincingness of an argument is the physical integrity of
a building best characterizes the metaphor at its appropriate level of schematic-
ity. (Clausner & Croft, 1997, p. 260)
 Conceptual metaphor theory 175

In other words, without establishing the appropriate level of schematicity, it is not

possible to answer the question of which elements of the source domain are mapped
onto the target, and which ones are not. If the source and target domains are too
schematic, then elements of the source could be mapped that are not mapped in
reality, and if conceptual metaphors are not sufficiently schematic, then elements
that are mapped in reality could not be mapped onto the target.
What then is the appropriate level of schematicity for conceptual metaphors?
Zinken (n.d.) suggests that the claim of traditional CMT does not stand up to
scrutiny. This is the idea that the mappings between the source and the target
are at the superordinate level in the sense of Rosch. For instance, in Lakoff and
Johnsons view the mappings of the LOVE IS A JOURNEY metaphor occur at the
level of VEHICLE, and not at the level of SHIP or TRAIN. In the metaphor the
mapping is between RELATIONSHIP and VEHICLE, and not between SHIP and
RELATIONSHIP or TRAIN and RELATIONSHIP. However, according to Zinken
in natural discourse the mappings occur not on the superordinate or subordinate
levels but between basic-level concepts. He argues that if they were on the super-
ordinate level, the basic-level concepts belonging to the same source domain that
have similar meanings would merely be alternative manifestations of high-level
mappings (and as such would have to be interchangeable), since the same mapping
would apply to them. However, Zinken suggests some apparent examples where
this is not the case: way and course belonging to the domain of PATH, ship and
boat to that of TRANSPORTATION, and kettle and pot to that of CONTAINER.
Zinken finds that in German and English political discourse speakers use these
words that belong to the same target domain and that have similar meanings in
rather different ways.
How is this possible if the words are linguistic instances of the same generic-
level mapping and they have similar meanings? In my view, to account for this, we
need two conceptual tools that are not part of the standard Lakoff-Johnson theory:
the scope of the source and main meaning focus (Kvecses, 2002, 2005). The scope
of source means that each metaphorical source domain applies to a certain group,
or set, of target domains. For example, the scope of BUILDING as a source domain
applies to any member of the target domain of COMPLEX ABSTRACT SYSTEM,
be it a political or economic system, human relationships, life, mind, and the like
(Kvecses, 2002). The main meaning focus means that each source domain is used
to conceptualize one or a small number of aspects of the target domain (Kvecses,
2002). For example, one of the meaning foci associated with the source domain
of BUILDING is that of structure. This can be given in the form of a (sub)meta-
phor: A COMPLEX ABSTRACT SYSTEM IS A COMPLEX PHYSICAL OBJECT
(Kvecses, 2002). That is, the use of the source domain of BUILDING allows us to
176 Zoltn Kvecses

conceptualize (to think and talk about) the structure-related aspects of complex
abstract systems.
With the help of these two conceptual tools, we can handle Zinkens counter-
examples as follows: The source domain of CONTAINER has as one of its meaning
foci tension, or pressure, yielding the (sub)metaphor: ABSTRACT TENSION IS
PHYSICAL PRESSURE. To express this meaning focus, the concept of KETTLE
seems more appropriate than that of POT because a kettle has steam inside it, gen-
erates pressure, and, as a result, even whistles, whereas the pot is mainly used for
storing and boiling a large quantity of water. Although we heat water in both (and
in this respect they have similar meanings), their main meaning focus and, conse-
quently, their metaphorical use will considerably and systematically be different.
Next, let us take the concepts WAY and COURSE. Zinkens argument would
be valid only if WAY and COURSE were examples of the same mapping. Though
the meanings of the two concepts are indeed very similar, they are best regarded as
examples of two different mappings. The concept of WAY belongs to the metaphor
that can be given as MEANS (OF ACTION) ARE PATHS, whereas COURSE is
an instance of the metaphorical mapping SCHEDULING HOW TO ACHIEVE
ONES PURPOSE IS SCHEDULING HOW TO REACH ONES DESTINATION,
in which the word course means a preplanned route to the destination.
Zinkens third example has to do with the words ship and boat. As he points
out, the two words are used very differently in their metaphorical senses in Ger-
man and English political discourse. The word ship typically indicates a state or
party in its metaphorical uses, while boat generally refers to the need for people
and parties to cooperate. Again, the question is whether the metaphorical uses of
the two words with similar meanings are based on the same mapping. They are
clearly not. In the case of boat we have the STATES ARE CONTAINERS meta-
phor, whereas in the case of ship it is two mappings that motivate the use of the
word. On the one hand, it is the LONG-TERM PURPOSEFUL ACTIVITIES ARE
LONG JOURNEYS, and, on the other, it is the COMPLEX ABSTRACT SYSTEMS
ARE COMPLEX PHYSICAL OBJECTS. Neither of these would be possible bases
for the metaphorical use of the concept of BOAT in the sense of state or political
party.
In sum, then, although Zinken is right in claiming that we cannot use words
with similar meanings metaphorically in the same way, this is not so for the reason
he suggests. Metaphorical mappings are found at the superordinate level, and the
different metaphorical uses of words with similar meanings is due to the meaning
foci associated with particular source domains, as well as to the fact that the words
are based on different mappings.
 Conceptual metaphor theory 177

IV. The issue of embodiment

Embodiment is one of the key ideas of cognitive linguistics that clearly distin-
guishes the cognitive linguistic conception of meaning from that of other cog-
nitively-oriented theories. In the emergence of meaning, that is, in the process
of something becoming meaningful, the human body plays a distinguished role
(Johnson, 1987; Lakoff, 1987; Lakoff & Johnson, 1999; Gibbs, 2006). It is especially
what is known as image schemas that are crucial in this regard. Image schemas are
based on our most basic physical experiences and are inevitable in making sense
of the world around us.
However, several researchers have pointed out that aspects of the view of em-
bodiment may lead to contradictions within the theory (e.g., Alverson, 1994; Ra-
kova, 2002). It can be problematic that the theory of embodiment tries to account
simultaneously for universality and cultural specificity. Rakova emphasizes that a
theory that builds on image schemas and, in general, on the universality of essen-
tial physical experiences cannot in the same breath be a theory of cultural varia-
tion especially not if embodiment is conceived naturalistically. Here are some
quotes that indicate her position:
The thing is that reductionism and relativism are not supposed to go together. The
failure to balance these two tendencies is, I believe, the second drawback of the
philosophy of embodied realism. (Rakova, 2001, p. 228)
Thus, my claim is that experientialism is often relativism in the strong sense, and
that the supposed universality of directly meaningful concepts and kinesthetic
image schemas is not consistent with the idea of culturally defined conceptualiza-
tions. (Rakova, 2001, p. 228)
The claim that there are cognitively significant cultural differences in the con-
ceptualization of spatial relations is incompatible with the naturalistic stand that
follows from the theory of image schemas. (Rakova, 2001, p. 238)

Undoubtedly, the examples that Lakoff & Johnson provide (like the CONTAINER
schema) may give the impression that Lakoff & Johnson regard image schemas
and embodiment as universal experiences that make things (including language)
meaningful in a natural way, that is, in a way that suggests that the universality of
embodiment mechanically produces universal meanings.
In my view, we can refine and improve on this conception of the embodiment
of meaning, and, thus, we can meet the challenge of the criticism above. In order
to do that, we need to change the way we think about embodiment; we should not
see it as a homogeneous, monolithic factor. This is made possible by the idea that
embodiment consists of several components and that any of these can be singled
out and emphasized by different cultures (or, as a matter of fact, individuals within
178 Zoltn Kvecses

cultures). I termed this idea differential experiential focus in previous work (see
Kvecses, 2005).
Let us take as an example the kind of embodiment that makes our concepts
and words relating to anger meaningful in different cultures. According to physi-
ological studies, anger is accompanied by several physiological reactions, such as
increase in skin temperature, in respiration rate, blood pressure, and heart rate
(Ekman et al., 1983). These are universal physiological reactions that derive from
the human body and explain why we find the same generic-level conceptual meta-
phor in languages and cultures that are independent from each other (Kvecses,
2002).
At the same time, we can observe that the different languages and cultures
do not attend to the same physiological reactions associated with anger. While
in English and Hungarian a rise in body temperature and increase in blood pres-
sure receive equal attention, in Chinese the presence of PRESSURE seems to be
much more focal (Yu, 1998). Moreover, as Rosaldos (1980) work tells us, the main
physiological characteristic of anger among the Ilongot of New Guinea is an un-
differentiated and generalized state of physiological arousal. In other words, it
seems that different languages and cultures base their anger-concepts on different
components and levels of embodiment, thereby creating partly universal, partly
culture-specific concepts. This account is made possible by the process of differen-
tial experiential focus.
The phenomenon of differential experiential focus can also be observed his-
torically. Gevaert (2001, 2005) suggests that in historical corpora of the English
language the conceptualization of anger as HEAT was prominent between 850
and 950. (This can be established on the basis of the number of heat related anger
metaphors in the various historical periods.) Later, however, anger was concep-
tualized mostly as PRESSURE, and, beginning with the 14th century, HEAT and
PRESSURE jointly characterized the conceptualization of anger in English. The
well known metaphor ANGER IS A HOT FLUID IN A CONTAINER is the end
product of the process. Gevaert justifiably asks in this connection whether the
Lakoff-Johnson view of embodiment can be maintained in light of such findings.
After all, it would be unreasonable to propose that the physiological responses as-
sociated with anger change from one century to the next.
The idea of differential experiential focus can serve us again in responding
to this criticism (Kvecses, 2005). The embodiment of anger, as we saw above,
is complex and consists of several components. Of these, as a result of certain
cultural influences over the ages, different components may occupy central po-
sition in the metaphorical conceptualization of anger. In other words, the criti-
cism formulated by Gevaert would only be valid if we thought about embodiment
as a homogeneous and unchanging factor in how humans conceptualize various
 Conceptual metaphor theory 179

abstract concepts. But if we think of embodiment as a complex set of factors to

which speakers can apply differential experiential foci, we can resolve the dilemma
raised by Gevaert and others.

V. The issue of the relationship between metaphor and culture

However, not even the weaker and less mechanical notion of embodiment de-
scribed above can provide a general account of how culture shapes metaphorical
conceptualization. It is not clear what the more precise relationship is between the
process of embodiment leading to universal metaphors and that of local culture
leading to language- and culture-specific metaphors. More generally, the question
is whether CMT can simultaneously account for both the universal and culture-
specific aspects of metaphorical conceptualization. This was the general issue I
tried to raise and resolve in my book Metaphor in Culture. For lack of space in this
paper, I can only briefly outline and demonstrate through some examples a pos-
sible answer to this question.
Metaphorical conceptualization in natural situations occurs under two si-
multaneous pressures: the pressure of embodiment and the pressure of context.
Context is determined by local culture. This dual pressure essentially amounts to
our effort to be coherent both with the body and culture coherent both with
universal embodiment and the culture-specificity of local culture in the course of
metaphorical conceptualization. We can achieve this in some cases, but in others it
is either embodiment or cultural specificity that plays the more important role.
Context may be characterized by physical, social, cultural, discourse, etc. as-
pects, and it consists of such factors as the setting, topic, audience, and medi-
um, which can all influence metaphorical conceptualization. For example, Boers
(1999) showed that physical context may systematically shape the way we think
metaphorically. Boers studied the ECONOMY IS HEALTH metaphor in a ten-
year period, and found that the use of this metaphor is systematically more fre-
quent in the winter than in the summer. ECONOMY IS HEALTH is a potentially
universal metaphor whose use varies according to the physical context of meta-
phorical conceptualization.
Which metaphor is used in a particular situation does not only depend on
which (potentially) universal metaphor is available in connection with the given
target domain for the expression of a given meaning but also on the setting and
topic of the situation in which the metaphorical conceptualization takes place. Let
us take the following passages from a Hungarian newspaper (Kvecses, 2005):
180 Zoltn Kvecses

Levelet rt Sepp Blatter a Nemzetkzi Labdarg Szvetsg (FIFA) svjci elnke

az zsiai szvetsg (AFC) vezetinek, melyben elfogadhatatlannak minstette a
kontinens kldtteinek hrom httel ezeltti kivonulst a FIFA-kongresszusrl,
ugyanakkor meggrte, hogy megprbl segteni az AFC gondjainak megolds-
ban jelentette kedden a dpa nmet hrgynksg.
Nagyon elkesertett az nk viselkedse a Los Angeles-i kongresszusunkon.
nknek, mint a labdargshoz rt szakembereknek tudniuk kellett volna, hogy
az a csapat soha nem nyeri meg a mrkzst, amelyik a lefjs eltt levonul a
plyrl ll a levlben. (Zalai Hirlap [The Chronicle of Zala County], July 28,
1999)
Sepp Blatter, the Swiss president of the International Football Federation (FIFA),
wrote a letter to the leaders of the Asian Football Association (AFC), in which he
deemed unacceptable the behavior of the associations delegates three weeks ago
when they left the FIFA Congress prematurely. On the other hand, he promised
that he would try to help solve the problems with which AFC is struggling the
German news agency dpa reported.
I was bitterly disappointed by your behavior at our Congress held in Los An-
geles. You, as experts on football, should have known that the team that leaves the
field before the game is called off by the referee can never win the game states
the letter. (my translation, ZK)

The passages are about a FIFA Congress, where the Asian representatives left the
meeting prematurely because of their dissatisfaction with some of the decisions
of the Congress. It is the behavior of the Asian representatives that is conceptual-
ized metaphorically by Sepp Blatter, the president of FIFA. The target domain of
his conceptual metaphor is the FIFA Congress and the source domain is football
itself. We find this completely natural. Why? In all probability, it is because the
congress is about football. That is to say, the topic of the congress (football) of
the FIFA meeting (the target) influences the conceptualizer to select a particular
source domain (the game of football). This is a common way in which we select
metaphorical source domains in local contexts.
The selection of the metaphors we use may also depend on who we are, that
is, what our personal history is or what our long-lasting concerns or interests are.
The Letters to the Editor sections of newspapers often offer a glimpse into how
these factors can shape metaphorical conceptualization. In a Hungarian daily, one
of the readers sent the following letter to the editor before Hungarys joining of the
European Union:
Otthon vagyunk, otthon lehetnk Eurpban. Szent Istvn ta bekapcsoldtunk
ebbe a szellemi ramkrbe, s vltoz intenzitssal, de azta benne vagyunk
akkor is, ha klnfle erk idnknt, hosszabb-rvidebb ideig, megprbltak ki-
rngatni belle. (italics in the original; Magyar Nemzet, [Hungarian Nation] June
12, 1999)
 Conceptual metaphor theory 181

We are, we can be at home in Europe. Since Saint Stephen we have been integrat-
ed/ connected to this intellectual/ spiritual electric circuit, and with varying degrees
of intensity, but we have been in it even though various powers, for more or less
time, have tried to yank us out of it (my translation, ZK)

The EUROPEAN UNION AS AN ELECTRIC CIRCUIT metaphor is not a con-

ventional one. But, as we find out from the readers note, he is an electrical engi-
neer, and it was only natural for him to create and use this particular metaphor.
In general, we can observe that our profession, personal history, concerns and
interests all play a role in how we arrive at the most appropriate source domains
for target domains in a given naturally occurring situation.
As the examples above clearly show, metaphorical conceptualization does not
simply and merely utilize ready-made and/or universal metaphors. The pressure
of context is another inevitable component in the use of metaphors. Our effort to
be coherent with the local context may be an important tool in understanding the
use of metaphors in natural discourse. This aspect of metaphor use has so far re-
mained outside the interest and, indeed, the competence of traditional concep-
tual metaphor theory. With the help of the new conceptual tools briefly introduced
in this section, the study of these exciting cases of metaphor use may open up new
possibilities in our understanding of linguistic and cultural creativity within the
framework of CMT.

VI. Conclusions

In this paper, I responded to five kinds of criticism directed in recent years at

CMT. The criticisms involve issues concerning methodology, direction of analysis,
schematicity of metaphor, embodiment, and the relationship between metaphor
and culture.
First, as regards the methodological issue, I pointed out that the goal of meta-
phor research as formulated by the critics and the goal of metaphor research in
traditional CMT are on different levels. The goal of cognitive linguists working
within the original Lakoff-Johnson framework is to postulate certain conceptual
metaphors at the supraindividual level, whereas the critics major goal is to iden-
tify linguistic metaphors systematically at the individual level (and then arrive at
hypothetical conceptual metaphors). Both goals are valid but only on their respec-
tive levels of metaphor analysis. In addition, the different goals complement each
other.
Second, another group of criticisms is concerned with the direction of meta-
phor analysis. Those who study metaphors bottom-up entertain the assump-
tion that metaphors at the linguistic level are characterized by a great deal more
182 Zoltn Kvecses

irregularity than those who work top-down care to admit. I hope I managed to
show that this criticism is only partially valid because by means of detailed meta-
phoric mappings we can explain many subtle differences in meaning at the linguis-
tic level. I also suggested in this connection that quantitative metaphor analysis
needs to be supplemented by intuitive qualitative analysis if the goal of conceptual
metaphor analysis is to reveal the nature and structure of abstract concepts in as
much detail and depth as possible.
Third, I considered the issue of the level of schematicity of metaphors. Some
critics, like Zinken, claim that in real discourse metaphors can be found at the
basic level, and not at the superordinate level. He argues that if they were on the
superordinate level, then the words that belong to the same physical domain and
that have similar meanings would have to have the same metaphorical meaning.
But since this is not so, mappings occur not at the superordinate level but at the
more specific basic level. As we saw above, we can account for the phenomenon in
a different way if we assume that particular physical source domains may have dif-
ferent meaning foci (e.g., the CONTAINER domain may have the meaning foci of
pressure and quantity), and, therefore, the concepts belonging to these source
domains may participate in different mappings. The different meanings arise natu-
rally from this account, and they are not anomalies within the theory.
Fourth, the key idea of embodiment in cognitive linguistics also became the
target of criticism. I agree that the notion is indeed problematic if it is conceived
mechanically and as something monolithic in all cultures. Even though this con-
ception has some use in accounting for certain universal aspects of metaphorical
conceptualization, it is useless if we wish to explain cross-cultural and historical
variation. We can avoid this problem if we replace the mechanical conception of
embodiment with what I have called differential experiential focus.
Fifth, scholars, such as Rakova, criticize conceptual metaphor theory on the
grounds that it attempts to account for cultural variation with a device that it uses
for explaining universality. This issue has to do with metaphorical creativity across
cultures. It is clear that not even the weaker, less mechanical understanding of em-
bodiment can help us with this issue. I suggested that we can get around the prob-
lem if we think of metaphorical conceptualization as a process in which speakers
are under two competing pressures: the pressure of universal embodiment and
that of local context.
In conclusion, I attempted to outline a framework for the study of metaphor
that, in several respects, can be thought of as a version of, or an alternative to, the
standard, traditional Lakoff-Johnson view. With its help, it becomes possible to
make conceptual metaphor theory more flexible, open, and, in my view at least,
stronger. This stronger view eliminates some of the weaknesses of the standard the-
ory and allows us to respond successfully to some important recent challenges.
 Conceptual metaphor theory 183

References

Alverson, H. (1994). Semantics and Experience. Universal Metaphors of Time in Tnglish, Manda-
rin, Hindi, and Sesotho. Johns Hopkins University Press.
Boers, F. (1999). When a bodily source domain becomes prominent. In R. Gibbs & G. Steen
(Eds.), Metaphor in Cognitive Linguistics (4756). Amsterdam: John Benjamins.
Clausner, T. & W. Croft. (1997). Productivity and schematicity in metaphors. Cognitive Science,
21(3), 247282.
Dobrovolskij, D. & E. Piirainen. (2005). Figurative Language: Cross-cultural and Cross-linguistic
Perspective. Amsterdam: Elsevier.
Ekman, P., Levenson, R.W., and Friesen, W.V. (1983). Autonomic nervous system activity distin-
guishes among emotions. Science, 221, 12081210.
Gevaert, C. (2001). Anger in Old and Middle English: a hot topic? Belgian Essays on Language
and Literature, 89101.
Gevaert, C. (2005). The anger is heat question: detecting cultural influence on the conceptual-
ization of anger through diachronic corpus analysis. N. Delbecque, J. Van der Auwera, and
G. Geeraerts (Eds.), Perspectives on Variation: Sociolinguistic, Historical, Comparative (pp.
195208). Berlin: Mouton de Gruyter.
Kvecses, Z. (1995). American friendship: the scope of metaphor. Cognitive Linguistics, 6, 315
346.
Kvecses, Z. (2000). The scope of metaphor. In A. Barcelona (Ed.) Metaphor and Metonymy at
the Crossroads (pp. 7992). Berlin: Mouton.
Kvecses, Z. (2000/2003). Metaphor and Emotion. Cambridge: Cambridge University Press.
Kvecses, Z. (2002). Metaphor. A Practical Introduction. Oxford: Oxford University Press.
Kvecses, Z. (2005). Metaphor in Culture. Universality and Variation. Cambridge: Cambridge
University Press.
Kvecses, Z. (forthcoming). Methodological issues in conceptual metaphor theory. In S. Handl
& H-J. Schmid (Eds.), Windows to the Mind: Metaphor, metonymy and conceptual blending,
To be published in the series Cognitive Linguistics Research. Berlin/New York: Mouton de
Gruyter.
Lakoff, G. & M. Johnson. (1980). Metaphors We Live By. Chicago: University of Chicago Press.
Lakoff, G. & Z. Kvecses. (1987). The cognitive model of anger inherent in American English.
In D. Holland & N. Quinn (Eds.), Cultural Models in Language and Thought (pp. 195221).
Cambridge: Cambridge University Press.
Pragglejaz Group. (2007). MIP: A method for identifying metaphorically used words in dis-
course. Metaphor & Symbol, 22(1), 139.
Rakova, M. (2002). The philosophy of embodied realism: a high price to pay? Cognitive Linguis-
tics, 13(3), 215244.
Rosaldo, M.Z. (1980). Knowledge and Passion. Ilongot Notions of Self and Social Life. Cambridge:
Cambridge University Press.
Steen, G. (1999). From linguistic to conceptual metaphor in five steps. In R. Gibbs & G. Steen
(Eds.), Metaphor in Cognitive Linguistics (pp. 5777). Amsterdam: John Benjamins.
Stefanowitch, A. (2007). Words and their metaphors. A corpus-based approach. In A. Stefanow-
itch & S.Th. Gries (Eds.,) Corpus-based Approaches to Metaphor and Metonymy (pp. 63
105). Berlin: Mouton de Gruyter.
184 Zoltn Kvecses

Yu, N. (1998). The Contemporary Theory of Metaphor. A Perspective from Chinese. Amsterdam:
John Benjamins.
Zinken, J. (n.d.). Levels of schematicity in metaphor: basic level source concepts. http://www.
port.ac.uk/departments/academic/psychology/research/ecologycultureandcommunica-
tion/filetodownload,23361,en.fdf.

Authors address
Zoltn Kvecses
Department of American Studies
Etvs Lornd University
Rkczi t 5.
1088 Budapest
Hungary

About the author

Zoltn Kvecses, Professor of Linguistics in the Department of American Studies at Etvs
Lornd University, Budapest. His main research interests include the conceptualization of emo-
tions, the study of metaphor and idiomaticity, the relationship between language, mind and
culture, and American slang and American English. He is also working as a lexicographer, and
is the author of several Hungarian-English, English-Hungarian dictionaries. His most recent
books include Language, Mind, and Culture. A Practical Introduction. 2006, Oxford Univer-
sity Press, Metaphor in Culture. Universality and Variation. 2005, Cambridge University Press,
Metaphor. A Practical Introduction. 2002, Oxford University Press, American English. An In-
troduction. 2000, Peterborough, Canada: Broadview Press, and Metaphor and Emotion, 2000,
Cambridge University Press. He has taught at several American and European universities, in-
cluding the University of Nevada at Las Vegas, Rutgers University, University of Massachusetts
at Amherst, Hamburg University, Odense University, University of Granada, and the University
of California at Berkeley. He is currently working on the issue of what it means to understand a
culture from a cognitive point of view and on the notion of abstract concepts.