1
Cognitive Metaphor Theory Revisited∗∗∗∗
ESTHER ROMERO AND BELÉN SORIA
Abstract
This paper provides a framework which, being compatible with Lakoff and Johnson’s theory
(1980), allows a description of metaphoric verbal utterances. The development of this
theoretical expansion is encouraged by Lakoff and Johnson’s distinction between nonliteral
and literal metaphoric expressions, and by the fact that they do not provide an explanation of
the nonliteral metaphoric use of expressions as distinct from the literal metaphoric one. They
simply say that metaphoric expressions are nonliteral when they are parts that are not used
in our normal metaphoric concepts. This suggestion is included in our model, in which a
metaphoric utterance is identified when the speaker perceives both a contextual abnormality
and a conceptual contrast, and it is interpreted using, among other things, a pragmatic
process of mapping to derive subpropositional metaphoric provisional meanings. This
explanation of the metaphoric mechanism allows an explanation of the utterances in which
nonliteral metaphoric expressions intervene without having to resort to a previous literal
interpretation of these utterances.
∗
Finantial support for this research, which has been carried out in the Project ‘El significado y los procesos
pragmáticos primarios y secundarios’, has been provided by DGICYT, code number BFF2003-07141. An
earlier version of this paper was presented at the II Latin Meeting for Analytic Philosophy in Aix en Provence
(13-15 November 2003). We are grateful to the participants of this Meeting, especially to Pierre Livet, our
commentator in Aix, for their useful questions and remarks.
2
1. Introduction
In Metaphors We Live By, Lakoff and Johnson (1980) introduce a powerful theory for the
study of the role of metaphor in our ordinary conceptual system. This theory has been
developed by Johnson (1987), Lakoff (1987), and Lakoff and Turner (1989), and followed,
although not always in complete agreement, by many other authors such as Gibbs (1990,
1992), Way (1991), Steen (1994), and Kövecses (2002), to mention only some. Lakoff and
Johnson’s main concern is with why our everyday concepts are structured in one way rather
than another. In particular, it is with conceptual metaphors, that is, conventional concepts
elaborated metaphorically that may take public form in metaphoric expressions. But these
metaphoric expressions, which represent these conventional metaphoric concepts, do not
need to be interpreted metaphorically.
Our purpose in this paper is not so much to specify the achievements of cognitive
metaphor theory for the description of human cognition as to state some limitations of it for
the analysis of the metaphoric use of language or, in other words, for the analysis of
imaginative metaphoric utterances, which must be interpreted metaphorically. These
limitations are related to the identification and interpretation of metaphoric utterances. We
provide criteria for both of these, and this allows the characterisation of imaginative
metaphoric utterances which express the non-conventional part of a metaphoric concept.
Being compatible with cognitive metaphor theory, our account grants access to a description
of the metaphoric use of language. If we want an explicit model for the analysis of
metaphoric utterances in general, we need to amplify cognitive metaphor theory. This
expansion is also acknowledged in some recent works such as Maalej (2001: 108) who
argues that a pragmatic component should be added to the cognitive metaphor approach in
order to account for metaphor processing.
3
In the present framework, we argue that a metaphor is identified when the speaker
perceives both a contextual abnormality and a conceptual contrast, and this identification of
the metaphoric utterance triggers the metaphoric mechanism for its interpretation, it triggers a
pragmatic process of mapping to get provisional metaphoric meanings.
To achieve our purpose, we first recall some relevant features of cognitive metaphor
theory. In particular, we examine, in section 2, the relations between metaphoric concepts and
expressions from a cognitive metaphor perspective. This exposition will show that Lakoff and
Johnson’s distinction between literal and nonliteral interpretation does not parallel the
distinction between literal and metaphorical. In the later, the distinction is about concepts,
and there are expressions of metaphoric concepts that must be interpreted literally; while in
the former the distinction affects interpretation directly. The terminological distinctions used
by cognitive metaphor theorists are not the traditional ones; these theorists introduced a
terminological shift that should be clarified to avoid misunderstandings when we open a
discussion among different theories. This leads to two assumptions in cognitive metaphor
theory which are incompatible with the task of explaining the interpretation of nonliteral
metaphoric uses of language. In particular, some of Lakoff and Johnson’s followers argue
that all types of utterances that include metaphoric expressions are interpreted in the same
way and they reject any criterion for the identification of metaphoric utterances. Precisely to
deny these assumptions, in section 3 we show that if we want to study metaphor in language
we have to go from metaphoric expressions to metaphoric utterances, utterances that are
always nonliteral or imaginative. And finally, in section 4, we provide a model which, being
compatible with earlier versions of cognitive metaphor theory although not so much with
subsequent developments, allows the characterisation of the identification and interpretation
of metaphoric utterances as distinct from literal ones. We propose that if we become aware of
the lack of parallelism between the conceptual and linguistic levels and accept the distinction
4
literal/noliteral when applied to metaphors (a distinction that was used in earlier versions of
cognitive metaphor theory), our way to expand their theory is a natural way to explain
imaginative metaphoric uses of language.
2. Metaphoric Concepts and Expressions from a Cognitive Metaphor Perspective
The most basic assumption in cognitive metaphor theory is that there is a set of ordinary
metaphoric concepts –conceptual metaphors– around which we conceptualise the world. The
concepts that our ordinary conceptual system includes structure what we perceive, how we
get around in the world, and how we relate to other people. Most of them are metaphorical,
and their essence is understanding and experiencing partially one kind of thing or experience
in terms of another.
As we can see in Table 1, according to Lakoff and Johnson, three types of metaphoric
concepts can be distinguished, marginal metaphoric concepts (Lakoff and Johnson 1980: 55),
conventional metaphoric concepts (Lakoff and Johnson 1980: 4), and new metaphoric
concepts (Lakoff and Johnson 1980: 145). Metaphoric concepts take public form in
metaphoric expressions, also called “metaphors”. For Lakoff and Johnson,
Metaphors as linguistic expressions are possible precisely because there are metaphors in a
person’s conceptual system. Therefore, whenever in this book we speak of metaphors, such as
ARGUMENT IS WAR, it should be understood that metaphor means metaphorical concept. (Lakoff
and Johnson 1980: 6)
5
In this quotation, we can see that they are using the term “metaphor” to refer to both concepts
and expressions indistinctly. The different types of metaphoric concepts and their relations
with metaphoric expressions can be seen in Table 1.
Table 1. Metaphoric concepts and metaphoric expressions in Lakoff and Johnson’s
approach
Metaphoric Concepts or metaphors Metaphoric Expressions or metaphors
Marginal metaphoric concepts
or marginal metaphors
Dead metaphor
Literal dead
metaphor (such as
(4) below)
Literal
metaphor Literal live
metaphor (such as
(1) below)
Imaginative live
metaphor of type
(a) (such as (2)
below)
Conventional metaphoric concepts
or
Conceptual metaphors Live metaphor
Imaginative live
metaphor of type
(b) (such as (3)
below)
New metaphoric concepts or new
metaphors
Imaginative or
nonliteral
metaphor
Novel metaphor
Imaginative novel
metaphor (such as
(5) below)
Metaphoric concepts are expressed through terms that express explicitly the two concepts that
play a part in a metaphor, and are represented in small capital letters. Among these concepts
only conventional metaphoric concepts are systematically called “conceptual metaphors”.
Conceptual metaphors or conventional metaphoric concepts are normally used when we think
and the expressions that represent them in a conventional way are systematically used in the
everyday language; they are concepts that we usually and systematically conceptualise in
terms of others. An example of conceptual metaphor is the metaphoric concept THEORIES ARE
BUILDINGS, where we partially conceptualise THEORIES in terms of BUILDINGS. Marginal
metaphoric concepts are ‘relatively uninteresting’ (Lakoff and Johnson 1980: 54). They are
isolated and unsystematic cases that do not systematically interact with other metaphoric
6
concepts because so little of them is used. An example of these is A MOUNTAIN IS A PERSON
(Lakoff and Johnson 1980: 55). New metaphoric concepts constitute a new way of thinking,
they have the power to create a new reality, to alter the conceptual system. They are not part
of our conceptual system. An example of these concepts is THEORIES ARE PATRIARCHS.
With respect to expressions, they basically make two distinctions (Lakoff and Johnson
1980: 53-55): first, they distinguish between literal metaphors, and imaginative or nonliteral
metaphors; second, they distinguish among dead metaphors, live metaphors, and novel
metaphors.
The second distinction among metaphoric expressions parallels the one drawn for
metaphoric concepts. Dead metaphors are expressions of marginal metaphoric concepts, live
metaphors are expressions that arise from conventional metaphoric concepts, and novel
metaphors come from new metaphoric concepts. This parallelism, however, peters out when the
distinction literal-imaginative comes between these distinctions (see Table 1 above). The result
of blending the two distinctions for the metaphoric expressions produces a third one by which
Lakoff and Johnson distinguish among literal dead metaphor, literal live metaphor, imaginative
live metaphor (subtypes (a) or (b)), and imaginative novel metaphor.
Let us begin with the description of the metaphoric expressions that come from
conceptual metaphors. These expressions are called “live metaphors”, and there are two types:
literal and imaginative. Literal live metaphors are the used part of a conceptual metaphor, for
example, “foundations” in (1)
(1) The foundations of my theory are sure.
is one of the parts commonly used of the conceptual metaphor THEORIES ARE BUILDINGS (Lakoff
and Johnson 1980: 53) and, therefore, (1) is a case of literal live metaphor. By contrast,
7
imaginative live metaphors are either (a) instances of the unused part of a usual conceptual
metaphor as “thousands of little rooms” in (2)
(2) His theory has thousands of little rooms. (Lakoff and Johnson 1980: 53).
which is a case of an unused part of THEORIES ARE BUILDINGS, rooms are parts of a building
which are not normally used as part of the concept THEORY, or (b) extensions of the used part
of a conceptual metaphor, such as “bricks” in (3)
(3) These facts are the bricks of his theory. (Lakoff and Johnson 1980: 53).
which is an extension of one of the used parts of that metaphoric concept: “the outer shell”.
This is used to structure the concept of a theory, but that is not the case of the material used to
construct the outer shell, the bricks. Examples (1)-(3) are “metaphorical” because they have
some expressions that are metaphorically attributed to what we are talking about, to wit, a
theory; they have metaphoric expressions (in italics) within the conceptual metaphor
THEORIES ARE BUILDINGS. This metaphoric concept is conventional. The metaphoric
expressions of this metaphoric concept are foundations in (1), thousands of little rooms in (2),
and bricks in (3). These metaphoric expressions can be literal as in (1) or imaginative as in
(2) or (3). The latter are interpreted metaphorically while the former is not. Let us see this
proposal in some more detail.
Since Lakoff and Johnson’s purpose is the study of the metaphoric structure of the
conceptual system, they are interested in the study of conventional metaphoric concepts or
conceptual metaphors we live by. In addition, they say:
8
Since communication is based on the same conceptual system that we use in thinking and acting,
language is an important source of evidence for what that system is like. (Lakoff and Johnson
1980: 3)
Thus, their study of conventional metaphoric concepts is focused in the study of metaphoric
expressions that come from these concepts; it is focused in examples such as (1)-(3). These
metaphoric expressions are all live metaphors. Conceptual metaphors or conventional
metaphoric concepts may materialise in examples of what Lakoff and Johnson call “literal
metaphors” and in examples of what they label “imaginative metaphors”. In fact, they say
explicitly
literal expressions (‘He has constructed a theory’) and imaginative expressions (‘His theory is
covered with gargoyles’) can be instances of the same general metaphor (THEORIES ARE
BUILDINGS). (Lakoff and Johnson 1980: 53)
The ordinary expressions of conceptual metaphors such as “He has constructed a theory” and
(1) are literal expressions, they are literally interpreted. Imaginative metaphors such as “His
theory is covered with gargoyles” and (2)-(3) are nonliteral metaphors, metaphors that are
interpreted metaphorically. Thus, live metaphors can be either literal, as in (1), an expression
interpreted literally, or imaginative, as in (2)-(3), expressions interpreted nonliterally. It is in
this sense that Lakoff and Johnson argue that parts of conceptual metaphors such as TIME IS
MONEY, or THEORIES ARE BUILDINGS are not literal and that
9
only part of them is used to structure our normal concepts. Since they necessarily contain parts
that are not used in our normal concepts, they go beyond the realm of the literal. (Lakoff and
Johnson 1980: 53-54. Our emphasis in italics)
Saying that for a concept to be considered nonliteral it has to ‘contain parts that are not used in
our normal concepts’ entails that it must take public form in an unusual way. This entails that an
expression can be used in an unusual way when it represents a conceptual metaphor. So, in
Lakoff and Johnson’s approach, the nonliteral or imaginative live metaphor is possible only
when we have expressions used in unusual contexts, in a context in which these expressions are
not normally used.1
This is the way in which we know we are facing a metaphoric expression
that must be interpreted nonliterally. Moreover, this quotation shows that it is its potential to be
revealed as a case of imaginative metaphor that enables the conventional metaphoric concept to
go beyond realm of the literal. This is due to the fact that the metaphoric conceptualisation of
one concept in terms of another is always partial. The quotation also shows that it is the nonconventional
or unusual use that makes an expression become nonliteral, and this is the case
even when the metaphoric concept is conventional.
In addition to live metaphors, there are expressions that, being classified as literal
metaphors, are not a part of a metaphoric system. In this case, the metaphoric concept of which
they form a part is not used systematically in our language; it is marginal. Expressions such as
the one illustrated in (4)
(4) We reached the foot of the mountain
1
This is the motivation of what we call “contextual abnormality”, see section 4 in this article and also X
(1997/1998).
10
are isolated instances of marginal metaphoric concepts. “The foot of the mountain” is the only
expression used of the marginal metaphoric concept A MOUNTAIN IS A PERSON (Lakoff and
Johnson 1980: 55). These metaphors are marginal in our language and culture, their used part
may consist of only one conventionally fixed expression of the language. Lakoff and Johnson
argue that examples such as (4) are literal metaphors in so far as these expressions are
interpreted literally. If we consider examples such as (1) and (4), we can say that our
conventional ways of talking about some things presuppose either a conventional metaphoric
concept or a marginal metaphoric concept and we are hardly ever conscious of their use.
Nevertheless, both concepts can be extended in imaginative ways.
Finally, novel metaphors are nonliteral metaphors that come from new metaphoric
concepts; they represent a new way of thinking. They are not used systematically to restructure
part of our conceptual system. This is the case with the expression in italics in (5)
(5) Classical theories are patriarchs who father many children most of whom fight
incessantly. (Lakoff and Johnson 1980: 53)
The metaphoric concept called forth in (5), THEORIES ARE PATRIARCHS, is new, and (5) must
be interpreted metaphorically. Examples (4)-(5) are metaphorical too, they have expressions (in
italics) that are metaphorically attributed to what we are talking about: in (4) about mountains,
and in (5) about classical theories. These expressions are foot in (4), and patriarchs who father
many children in (5). The metaphoric concepts involved in these examples are different in nature
from the metaphoric concept involved in (1)-(3), THEORIES ARE BUILDINGS. Only the latter forms
a part of our conceptual system.
Within nonliteral metaphors we must not only include imaginative live metaphors but
also imaginative novel metaphors, expressions that must be interpreted metaphorically and
11
which do not depend on a conventional metaphoric concept. Lakoff and Johnson characterise
“imaginative novel metaphors” as:
Instances of novel metaphor, that is, a metaphor not used to structure part of our
conceptual system but as a new way of thinking about something. (Lakoff and Johnson
1980: 53)
Until now, we have expounded not only the features of the different types of
metaphoric concepts and their metaphoric expressions but also the relations between them.
This exposition, nevertheless, shows that cognitive metaphor theory presents several
problems. First, for Lakoff and Johnson, (1)-(5) are all examples of metaphors but (1) and (4)
must be interpreted literally while (2), (3) and (5) nonliterally. Thus, their distinction between
literal and nonliteral interpretation is not related to the distinction literal vs. metaphorical
even though it should be, at least, in a theory which intends to explain the metaphoric use of
language. The assumption that metaphors are metaphors because they give expression to a
metaphoric concept does not entail that this is useful to explain how we relate linguistic
utterances with interpretation, especially those which require a metaphoric or nonliteral
interpretation.
In addition, Lakoff and Johnson change the traditional senses associated with the
terminology as we can see in Table 2.
12
Table 2. The terminological shift in cognitive metaphor theory
Traditional terminology Cognitive metaphor terminology
Dead metaphor
Dead metaphors or conventional metaphors Literal metaphor
Live metaphor
Live metaphors or novel metaphors
Imaginative or nonliteral
metaphor
Novel metaphor
Traditionally (see Leech 1969; Ricoeur 1975; Searle 1979; or MacCormac 1985) it is considered
that, as far as expressions are concerned, dead metaphors are opposed to live metaphors.2
This
distinction is equivalent to the distinction in cognitive metaphor theory between literal and
imaginative metaphors. Traditional live metaphors correspond to Lakoff and Johnson’s
imaginative or nonliteral metaphors; traditionally dead metaphors are lexicalised metaphors and
correspond to literal metaphors in cognitive metaphor theory which means that they are called in
most of the occasions “live”. Lakoff and Johnson change the sense in which it is said that a
metaphor is “live”. From a traditional point of view, “live” refers to a non-conventional
metaphor, an expression which is not lexicalised. Indeed, this terminological change is clearly
appreciated in the following quotation:
Expressions like ‘wasting time’, ‘attacking positions’, ‘going our separate ways’, etc., are
reflections of systematic metaphoric concepts that structure our actions and thoughts. They are
‘alive’ in the most fundamental sense: they are metaphors we live by. The fact that they are
conventionally fixed within the lexicon of English makes them no less alive. (Lakoff and
Johnson 1980: 55)
2
The distinction between conventional and novel metaphor parallels this one (see, for example, Indurkhya 1992,
and XX 1998).
13
Live metaphoric expressions are, in the traditional terminology, nonliteral or nonconventional
expressions and, in the cognitive terminology, some of them are literal or
conventional expressions that come from metaphoric concepts we live by. The terminological
shift is possible because, in the former, it is the expression that is “live” and, in the latter, it is
we (and not the expression) that live by the metaphoric concepts. From the perspective of
cognitive metaphor theory, live metaphoric expressions give expression to conventional
metaphoric concepts either literally or nonliterally. They distinguish between dead, live and
novel metaphors; but although they use two of the labels traditionally used to distinguish
metaphor, those labels are not used in the traditional way. In the traditional terminology,
there are neither two different types of literal metaphors as in Lakoff and Johnson (dead and
live) nor two types of imaginative metaphors (live and novel). For authors outside the
cognitive frame, there are only dead (literal) metaphoric expressions or live (imaginative or
nonliteral) metaphoric expressions. This terminological clash has entailed an added difficulty
to the distinction between imaginative and literal metaphors.
Surely, the proposal that metaphors are metaphors because they give expression to a
metaphoric concept, together with the idiosyncratic variations in the terminology introduced
by these authors have generated problematic assumptions in the task of explaining such basic
aspects of metaphor as how to identify verbal metaphors and how to relate them to their
interpretation, an interpretation which must be metaphorical and thus nonliteral. Among these
problematic assumptions, we wish to highlight two. First, the view that all types of metaphor,
imaginative or not, are interpreted in the same way, or that all the utterances that involve
expressions of metaphoric concepts are interpreted by the same mechanism. Second, the
rejection of any criterion for the identification of imaginative metaphoric utterances. These
assumptions contrast with our position since we find it right to speak about imaginative
metaphor as a different phenomenon from literal metaphor. This, in our opinion, coincides
14
with Lakoff and Johnson’s former proposal (Romero and Soria 1997/1998). In Lakoff and
Johnson (1980) there is no account of how to identify or interpret metaphoric utterances, but
they recognise that some metaphors are parts not used of our concepts and that these parts
must be interpreted nonliterally.
Although in section 4 we will see in what sense the assumptions mentioned are not
acceptable, we need first to point out that some adjustments are required so that the
terminological clash between theories is overcome and we can avoid misunderstandings.
3. From Metaphoric Expressions to Metaphoric utterances
In our opinion, if we are to talk about interpretation, we have to make various changes to this
view. First, we must speak about utterances and not about expressions. Theoreticians from a
cognitive metaphor perspective normally talk about concepts or expressions, but we want to
characterise utterances as, in our opinion, these are the only interpretable units. If we want to
provide a model as an aid to an explanation about the interpretation of certain metaphoric uses of
expressions, we must account for utterances and not for the concepts that the reader has in his
mind or for expressions that are out of context. The relevant unit for assigning meaning is the
utterance, and not the sentence or any other expression out of context.
In this sense, a second change in cognitive metaphor theory must be introduced: we
have to establish the distinction between different types of metaphors as regards the
interpretation of language in use. Thus, the distinction between literal and nonliteral can and
must be made and the features that allow drawing the line between them form an explicit part
of the model that will allow us to analyse all types of metaphoric utterances. If we want an
explicit model for the analysis of metaphoric utterances in general, we need to amplify the
model provided by cognitive metaphor theory. We have to account for metaphoric utterances,
15
utterances that are interpreted nonliterally. As our interest is not to describe concepts but
metaphoric utterances, any use of what Lakoff and Johnson call “literal metaphors”, which
are not metaphoric utterances but literal expressions, will not be the focus of our approach.
An utterance of these expressions in a normal context will be a literal utterance. A metaphoric
utterance is possible only if an expression is used in an imaginative way, that is, in an
abnormal context that leads us to contrast two concepts in a new way. Metaphoric utterances
are always nonliteral.
A model to account for metaphoric utterances must include the features that
characterise imaginative metaphor, features of metaphoric identification and interpretation
that will be explained later, and with which we would be able to characterise the metaphoric
use of language when it involves both the non-conventional part of conceptual metaphors,
and new metaphoric concepts. Nevertheless, before facing this aim we consider it necessary
to expound the general sense in which conceptual system will be used here.
A competent speaker knows the conventional codes that shape communication. By
the linguistic competence of individuals we understand the phonetic, lexicogrammatical and
semantic conventions shared by the members of a linguistic community in a certain stage of
the language. Our use of the term “linguistic competence” differs from the Chomskyan sense.
We add “of individuals” to underline the fact that although the code is the central matter in
this notion, it must refer to the conceptual system of speakers. “(…) individuals become
habituated to a code-like predictability of usage, forms, and meanings” (Toolan 1996: 9)
which it is dependant on their linguistic and non-linguistic past experiences. Indeed, the
conceptual system of a competent speaker includes the individual’s encyclopaedic knowledge
(Sperber and Wilson 1986: 142) and the context of culture. In addition, it involves the
mechanisms of interpretation of possible utterances and the contextual potential of the lexical
units comprised in the vocabulary of a language. This means that, depending on our linguistic
16
competence, there is some habitual or expected occurrence of words or concepts in context that
is characteristic of semantic or lexical behaviour in language, testifying to its predictability of
use.
In communication, people need a mutual cognitive environment to understand each other
successfully (Sperber and Wilson 1986). The conventional meanings of the words in a language
are the community’s store of established knowledge. This knowledge is achieved by experience
(linguistic or extralinguistic). Nevertheless, we would like to point out that language also has
some resources capable of breaking the stability. One of them is the metaphoric use of language;
a fresh use of language must be interpreted metaphorically.
The imaginative use of language shows a provisional metaphoric restructuring of a
concept (Romero and Soria forthcoming(a)). This provisional metaphoric restructuring of a
concept starts from conventional meanings to create metaphoric provisional meanings which do
not belong in the code but for which the code is its condition of possibility. This is the kind of
meaning Lakoff and Johnson (1980: 139) took into account in their chapter ‘New Meaning’, and
yet, this has been ignored by many of their followers who have even argued that there is no
difference between the literal and the metaphoric interpretation. In addition, the concepts these
meanings represent do not need to be included in the conceptual system, they are not concepts
we live by or concepts we will live by in the future. Provisional metaphoric concepts permit
generating notions that were not available before and may not become conventional and, in this
sense, not only do they reflect a certain world-view but also allow modifying it provisionally.
Imaginative metaphor can break conventional coding and promote a new or fresh relation
between language and meanings. These and other characteristics of imaginative metaphor will
be considered in the next section of this paper.
17
4. Some Characteristics of Imaginative Metaphoric utterances
From our point of view, unlike literal utterances, metaphoric utterances are identified when
the speaker perceives both a contextual abnormality and a conceptual contrast. These are the
two conditions for the identification of the metaphoric use of language. Metaphoric
interpretation is achieved when the speaker applies, among other things, the metaphoric
mechanism; a mapping mechanism between two domains so that we can construct a
provisional metaphoric concept that links the two domains (Romero and Soria forthcoming
(a)). Unlike what happens with the application of the literal mechanism, through the
application of the metaphoric mechanism the addressee obtains at least a metaphoric
provisional meaning. But let us see what this means resorting to examples of both literal and
metaphoric uses of language.
To identify a certain use of language as literal, we have to resort to, in the same way as to
identify the metaphoric use of language, both the linguistic competence of the interlocutors and
the context in which the language is used. In the linguistic competence of individuals, possible
contexts of use, linguistic or extralinguistic, are predicted for lexical items. The literal use of
language is characterised as that in which linguistic expressions appear in normal linguistic and
extralinguistic contexts. If nothing indicates that the utterance must be interpreted in a different
way, it will be identified as literal. Let’s take (6)
(6) [Sarah asks Marian where her pet is and she answers:] My cat is on the mat.
as an example of literal utterance. As an utterance, this example includes part of its context
which is represented in square brackets. Although the context of interpretation would be much
richer, we find it important to include some relevant features about the context and assume that
18
the rest would be the most normal and available ones without further specification.
In (6) Sarah perceives that both the linguistic and the extralinguistic context in which
every word included in the sentence “My cat is on the mat” appears coincide with one of the
potential contexts fixed for them in the linguistic competence of the speaker. If a speaker
does not transgress any of the norms in which the combinatorial potential of lexical items
take shape when used, a literal use of language is made. The same can be said about a usual
utterance of (1), which is an utterance of what Lakoff and Johnson call “literal metaphor”.
The linguistic context in which every word included in the sentence “The foundations of my
theory are sure” coincides with one of the potential contexts fixed for them in the linguistic
competence of the speakers. With a usual utterance of (1), a literal use of language is made.
Thus, the usual utterances of (1) must be called “literal utterances” and must be interpreted
literally. Conventional metaphors are, in our opinion, interpreted literally. This is something
maintained also by the subsequent development of cognitive metaphor theory (see Lakoff and
Turner 1989; Way 1991; Keysar and Gluksberg 1992; and Gibbs 1994). But we say so because
we think that they have become literal, not because we argue, as they do, that metaphors (in
general) are interpreted in the same way as literal utterances. We think that literal and genuine
metaphoric meanings are of a different nature: the former are conventional and interpreted
literally; the latter are provisional and interpreted metaphorically. Only, in some cases, the
metaphoric meaning begins a process of conventionalisation. It is after this process that the use
of an expression ceases to be genuinely metaphorical and thus a metaphoric interpretation is not
required. Many of the concepts that conform our conceptual system have been structured
metaphorically, but with time they have passed to form a part of our conventional
conceptualisation of the world. Thus, when our utterances make reference to these concepts, the
utterances are not metaphorical. From a synchronic point of view, they should be described as
literal utterances (Romero and Soria 1998).
19
If a speaker perceives both a contextual abnormality and a conceptual contrast in an
utterance, s/he will think that the use of language is metaphorical. A contextual abnormality is
produced when an expression is used in an unusual linguistic or extralinguistic context and a
conceptual contrast is brought about when we identify one concept as source domain and
another concept as target domain. If Marian utters “My cat is on the mat” as an answer to
Sarah’s question about the location of her one-year-old son, Sarah clearly perceives that it is an
anomalous way of talking about Marian’s son. The contextual abnormality in (7)
(7) [Sarah asks Marian where her one-year-old son is and she answers:] My cat is on the
mat
is produced by the occurrence of a lexical item in an unusual extralinguistic context. In (7) we
perceive a contextual abnormality, as infants are not the kind of thing that we call “cats”
according to our conceptual system. The possible occurrences of lexical items, such as “cat”, in
their different contexts of use makes up the conceptual system that shapes our linguistic
competence. If the context does not fit with the prediction of use of the concepts fixed in our
linguistic competence, we will be facing an unusual context that will lead us to think that the
utterance may be nonliteral. Abnormality (or anomaly or deviation or incongruity) has been
used by other authors such as Kittay (1987), Ortony (1980), or Bailin (1999) as the main
criterion for the identification of metaphor, but we disagree because we do not describe the
criterion in the same way (abnormality is not described here as a literal falsehood or as a
semantic deviance, rather it is characterised as a case of contextual departure from
conventional language usage) and we think that, although it is a necessary condition for the
identification of imaginative metaphors, it is not sufficient.
20
Abnormality does not suffice; it does not even serve to delimit the nonliteral use of
language from others; abnormality may be found in non-sense utterances, fictional utterances
and in metonymy (Romero and Soria 2003). Thus, to delimit them we have to find an additional
criterion, which in metaphor is the conceptual contrast. The abnormality in (7) leads us to
recognise that two concepts are involved and that one of them acts as the target concept, the
concept we are talking about or the concept INFANT, and the other one as the source concept,
the concept attributed to the one we are talking about or the concept CAT, taking into account
that among the senses of “cat”, the one that serves to classify the infant nonliterally is the
sense of “cat” in which it is conceived as “a small domesticated feline mammal”. Now we can
say we are facing a metaphoric utterance, and metaphoric interpretation is required; a pragmatic
process of mapping is needed to obtain metaphoric interpretation which is not in literal
interpretation.
Even though the context is always involved and helps to determine what type of
interpretation we face, the role of context does not become exhausted in this task. It is nowadays
a commonplace in all theories of meaning to admit that context is essential for the processing of
utterance interpretation. Where they disagree is in the proposals that specify its function or
functions. The characterisation of the different types of pragmatic processes involved in
utterance interpretation is an open task that we will not tackle here for obvious reasons of
space (see Sperber and Wilson 1986; Bach 1994; Wilson and Sperber 2002; Carston 2002;
Recanati 2004; and Romero and Soria forthcoming(b)). In what follows, we will illustrate the
type of proposals on the role of context in interpretation that we defend.
If we consider (6) again, we identify a literal use of language that leads us to interpret it
literally, to construct the literal informative content that (6) expresses. The speaker’s meaning of
(6) is specified indicating, among other things, what the speaker says with (6) which, in its turn,
depends on both the linguistic meaning of the constituents of the sentential expression contained
21
in (6) and the syntactic-semantic rules that allow their co-occurrence and which are consistent
with the compositional principle. The most basic constituents of the literal content are, in the
same way as those of the metaphorical, the conventional meanings of the lexical items that are
involved in an utterance. These meanings are the ones available in the linguistic competence of
speakers which, as we said, predict possible contexts of use for the lexical items that are learned
and stored linked to contexts. The possible contexts associated to a lexical item form a part of
the conceptual system of the individual.
The content that (6) expresses will depend to a great extent, and seemingly will (7), on
the conventional meaning of “my”, “cat”, “be”, “on”, “the” and “mat”. Some of these items are
far from having only one conventional meaning. They are ambiguous. It is context, linguistic
and extralinguistic, that will allow us to choose one of the possible conventional meanings of
such terms, we select contextually one of the senses as appropriate. In (6), the conventional
meaning of “cat” that we contextually select is that of “a small domesticated feline mammal”
dismissing the other senses of this term (e.g. “a woman who gossips maliciously”, or “nautical.
a heavy tackle for hoisting an anchor to the cathead”). We choose this sense of “cat” because the
extralinguistic context confirms that we are talking about a feline, about Marian’s pet.
Nevertheless, the semantic value contextually assigned to a sentence constituent in the
process of utterance interpretation is not always merely conventional. Thus, on many occasions
we are forced to enrich contextually the meaning of constituents (Sperber and Wilson 1986:
188-191). “My” functions as determiner of “cat” which is the head of the noun phrase “my cat”.
This phrase is an expression that presents referential indeterminacy to the extent that we do not
know what cat we are talking about until we know who the speaker of the utterance is. In this
case, when Marian says “my cat” she is talking about her cat, but on other occasions we could
be talking about another person’s cat by using this expression. At this point, another semantic
indeterminacy that we normally eliminate by resorting to context arises, to wit, the
22
indeterminacy of the type of relation between the speaker and the cat. In a fictional situation, a
cat could speak about her offspring by uttering this expression, and the relation between the cat
speaking and her offspring would be, in this case, one of progeniture, while in (6) a possessive
relation is involved if we accept as a background assumption that Sarah and Marian are human.
Any of these uses of “my cat” is a literal use of the expression and they all stem from the same
conventional meaning, although–depending on the cases–the object they denote and the
possessive relation the speaker has with that object may vary.
The noun phrase “my cat”, that functions as subject in (6), is followed by two other
phrases, the verb phrase “is” and the prepositional phrase “on the mat”, whose meanings
depend, in their turn, on the conventional meanings of their constituents and, if it is required, on
context. What these phrases mean is combined again until we build what is said with (6). Once
we have resorted to compositionality it seems that the process of elaboration of literal meanings
is clearly contextual.
But it is not always the case that the semantic value contextually assigned to the
constituent is conventional or depends on the enrichment of a conventional meaning;
sometimes it is a transformation of one of the conventional meanings, enriched or not, that
substitutes for this meaning. We are thinking now about the semantic value that is
contextually acquired by some terms uttered in their nonliteral uses such as the metaphorical.
The difference between the metaphoric and the literal meanings of utterances can be
drawn as follows. The literal is a form of expression in which one of the conventional
meanings of every lexical unit (or even part of it) is operative until the end of the process of
interpretation. In the metaphoric use, once one of the conventional meanings of the most
basic components is established, we apply a set of procedures that allows us to elaborate
some metaphoric provisional meaning; we elaborate a metaphoric provisional meaning at
least for those components that constitute the vehicle of metaphor. The metaphoric meaning
23
of the utterances is constructed not only by means of the procedures of literal interpretation
but also by means of some procedures of meaning of a different nature.
Once the speaker has recognised that the utterance of (7) is a case of metaphor, s/he
interprets it and, doing so is but applying the mechanism that characterises this phenomenon.
In particular, it is to link two separate cognitive domains, by using the language appropriate
to one of them as a lens through which to observe the other. To put it technically, it is to
apply a pragmatic process of mapping from the source domain to the target domain so that we
can construct a subpropositional provisional meaning. Metaphors redescribe a domain called
the “target domain” in terms of another domain called the “source domain”, selecting,
highlighting, omitting, and organizing certain features of the target domain (Black 1954;
Indurkhya 1986; Gineste, Indurkhya and Scart 2000). In addition, often some features will be
created within the target domain. This is coherent with Lakoff and Johnson’s idea, pointed
out above, that the essence of metaphor is understanding and experiencing partially one kind
of thing or experience in terms of another. The target domain represents the concept we are
talking about, and the source domain represents the concept attributed to the one we are
talking about. The domains represent the conventional conceptions of the two concepts that we
detect in the conceptual contrast, and they consist of a set of terms which make up its
vocabulary and a set of sentences which specify how these terms are related to the information
associated with the concept.
In (7) a metaphoric use of language is identified and, thus, at least its metaphoric
vehicles -the terms metaphorically attributed to what we are talking about- must be interpreted
metaphorically. In (7) the vehicles are the terms that are involved in the description of CAT.
Thus, the relevant question is: what is the meaning of “cat” in (7)? The extralinguistic context
alludes to a child, Marian’s son. We are not talking about a cat in any of the conventional senses
of the term “cat” and, among the senses of “cat”, the one that serves to classify the infant
24
nonliterally is the sense of “cat” that also intervened in the interpretation of (6), it is the one
identified as source domain.
To create the metaphoric provisional meanings we must establish what the metaphoric
relation of the concepts involved in (7) is; we must establish the metaphoric restructuring or
recategorisation. This is done by mapping a set of sentences from source domain, cat, to target
domain, infant. We may coherently transform a set of sentences of the source domain to
sentences with terms only of the target domain and this set of transformed sentences will
redescribe the concept INFANT through the concept CAT. The result of the metaphoric
restructuring, the restructured target domain, provides us with a context of interpretation of the
metaphoric utterance provisionally restructured for that occasion. The mapping needed to
interpret (7) generates a metaphorically restructured target domain, a conception of infants
provisionally modified by those aspects of the concept CAT that influence its restructuring. When
the context from which the terms that are involved in a metaphoric utterance changes, the
meanings associated with these terms change provisionally. In this way, some terms acquire a
metaphoric meaning.3
The relation between the terms and the originated metaphoric meanings is
not established or conventionalised. Hence, in (7) terms such as “cat” acquire a metaphoric
provisional meaning. Such meanings are provisional because at the moment that the metaphoric
utterance disappears, they also vanish (Romero 1990/1991). The metaphoric relation of the
concepts involved in (7) consists in mapping a subset of information associated with the
source domain, information about the typical features of cats, into the set of information
associated with the target domain; in particular, mapping the piece of information that is
coherent with the description of the target domain, with the concept of infant. The
information mapped is the information that, together with the target domain, composes the
metaphoric provisional meaning of “cat”. The meaning of “cat” is the meaning that this term
3
Some authors such as Davidson (1978) and Cooper (1986) deny the existence of metaphoric meaning.
25
has in the metaphorically restructured target domain, it is the information that can describe
INFANTS coherently when infants are seen as bearing the features typical of cats; it is a
question of seeing an infant as a being that needs feeding and care, goes on all fours, plays
with anything available, scrutinises things carefully and is unfriendly with the unknown.
“Cat” acquires a transferred metaphoric provisional meaning.
This information constitutes the context from which that utterance is interpreted. Thus,
this is what allows us to establish what is said through (7). Since “cat” means here provisionally
and metaphorically what we have just specified, the infant with the typical features of cats, “my
cat” in (7) denotes Marian’s son and Marian is the progenitor of the infant that the speaker is
denoting when she uses “cat” metaphorically. Once all these changes have been made, what is
said with (7) is composed in the same way as what is said with (6). What varies is a process of
contextual interpretation that intervenes at the level of constituents, the process of interpretation
characteristic of metaphoric interpretation. What is obvious here is that the speaker does not
have the literal interpretation of the sentence included in (7) at any moment of the interpretation.
There are not two stages in metaphoric interpretation. When we say that Marian’s son is
her cat it does not mean that this infant is literally a cat but that this is false; it means that “cat”
acquires here a transferred meaning as a result of the mapping from the domain representing the
concept CAT to the domain representing the concept INFANT. The new recategorisation produced
in this imaginative metaphor by the mapping from a source domain to a target domain shows
that metaphor is a mechanism used for producing metaphoric provisional meanings not available
in the system of the language. In metaphor old words are used for new jobs (Goodman 1968); it
constitutes a cost-effective, practical and creative way of using terms. Metaphoric utterances,
regardless of the field in which they are produced, yield a recategorisation of the concept
metaphorically restructured.
26
Let us see how this recategorisation takes place in another imaginative metaphoric
utterance. In an utterance of the expression represented in (5) such as (8)
(8) [A professor talking about Chomsky’s theoretical developments, says:] Classical theories
are patriarchs who father many children most of whom fight incessantly
we recognise the anomalous or abnormal use of “patriarchs” as theories, since they cannot be
classified as patriarchs in our conventional conceptual system. The context in which
“patriarchs” appears is not predicted as one of the normal contexts fixed for the use of the
concept PATRIARCH in our linguistic competence. Thus, we are facing an unusual context that
may lead us to a possible conceptual contrast. In (8) we know, by the context, that the topic
of the conversation, what we are talking about (the target domain) are theories and this,
together with the abnormal use of the concept PATRIARCH leads the hearer to identify the
latter as the source domain. Once the hearer has realised that PATRIARCH is used as the
concept represented by the source domain and the concept THEORIES as the one represented
by the target domain, the utterance is identified as metaphorical and a metaphoric
interpretation is required.
But this does not mean that we have to interpret an utterance of the expression “Classical
theories are patriarchs who father many children most of whom fight incessantly” in two stages:
first literally and then, metaphorically. Quite on the contrary, we think that a literal interpretation
of (8) is not available. The utterance is interpreted directly, though not necessarily in the same
way as the literal-conventional use of language. The denial of metaphoric interpretation in two
stages does not necessarily entail that metaphoric and literal (or conventional) interpretations are
achieved in the same way. To interpret metaphoric uses of language, we have to map certain
features from one domain to the other in order to elaborate a metaphoric provisional meaning at
27
least for those elements that constitute the vehicle of the metaphor. In (8), these elements are the
subpropositional constituents that are conventionally included in the source domain, patriarch,
and attributed to the target domain. To achieve these metaphoric nonconventional meanings we
have to apply a pragmatic process of mapping to establish what the metaphoric relation of the
concepts involved in (8) is. This relation is established by mapping the piece of information of
the description of the source domain, patriarch, that is coherent with the concept of theory. By
means of this mapping, the concept THEORY is redescribed through the concept PATRIARCH and
as a result we obtain a provisionally restructured target domain which provides us with a context
of interpretation of the metaphoric utterance. The metaphorically restructured target domain is,
in this occasion, a conception of theories provisionally modified by those aspects of the concept
PATRIARCH that influence its restructuring. The context from which the interlocutor interprets the
terms that are involved in the metaphoric utterance (8) is different from any of the normal
contexts in which they might be used and thus the meanings associated with these terms change
provisionally. In this way, some terms such as “patriarchs”, “father”, “children” acquire
metaphoric meanings and the relation between these terms and the originated metaphoric
meanings is not established or conventional.
The meanings of “patriarchs”, “father”, and “children” are the meanings that these terms
have in the metaphorically restructured target domain and they acquire a transferred metaphoric
provisional meaning. The meaning, for example, of “patriarchs” is the information that can
describe THEORIES coherently when theories are seen as bearing the features typical of
patriarchs; it is a matter of seeing a theory which originated a set of other related theories as the
head of a tribe-like set of theories, and as such the latter can be mutually antagonistic although
sharing some of the features inherited from the same theory.
Unlike what happens in the description of nonliteral metaphoric interpretation, when we
describe and interpret examples of literal metaphoric expressions such as “foundations” in (1)
28
we do not have to resort to the kind of analysis elaborated for (7) or (8). It is not abnormal to use
the term “foundations” when talking about theories; we do not have a target domain and a
source domain, and so we do not have to map any feature from one domain to the other. We can
interpret a normal utterance of this expression literally without the need to contrast two domains.
The concept THEORY is structured with expressions such as “construct”, “foundation”, etc.
whose meanings depend on the way in which the metaphoric concept THEORIES ARE BUILDINGS
is used to structure the concept THEORY.
5. Conclusion
A linguistic analysis guided by cognitive metaphor theory only specifies what the metaphoric
concepts are that literal and imaginative metaphors come from, but from the point of view of
identification and interpretation, this approach does not provide a key to metaphor classification
keys. It is a theoretical mystery to know when and how expressions are used nonliterally. Thus,
cognitive metaphor theory provides us with a limited account of the metaphoric use of language.
If the expression used does not entail a conventional use of a metaphoric concept, how do we
know that an expression constitutes a non-conventional use of a metaphoric concept? We need
criteria not only of identification but also of interpretation. It is necessary to characterise the
metaphoric meaning created for the occasion. Contextual abnormality and conceptual contrast
are the joint criteria proposed here to identify an utterance as metaphorical, while mapping is the
pragmatic process to produce subpropositional metaphoric provisional meanings that take a part
in what is said with a metaphoric utterance. In this article, we have developed this aspect about
“new meanings” which is overlooked in the theory of cognitive metaphor, showing the
characteristics of imaginative metaphor, the characteristics of metaphoric utterances. Even if we
were interested in characterizing conceptual metaphors, as is the case in cognitive metaphor
29
theory, we would have to resort to an explanation such as the one offered here, at least, to give
form to the imaginative part of these conventional concepts.
University of Granada
30
References
Bach, Kent (1994). Conversational Impliciture. Mind and Language 9: 124-62.
Bailin, Alan (1999). No Man Is An Island: Negation, Presupposition and The Semantics of
Metaphor. Journal of Literary Semantics XXVIII/1: 58-75.
Black, Max (1954). Metaphor. Proceedings of the Aristotelian Society 55: 273-94.
Carston, Robin (2002). Thoughts and Utterances. The Pragmatics of Explicit Communication.
Oxford: Blackwell.
Cooper, David (1986). Metaphor. Oxford and New York: Basil Blackwell.
Davidson, Donald (1978). What Metaphors Mean. Critical Inquiry 5: 31-47.
Gibbs, Raymond W. (1990). The Process of Understanding Literary Metaphor. The Journal of
Literary Semantics XIX/2: 65-79.
Gibbs, Raymond W. (1992). When is Metaphor? The idea of Understanding in Theories of
metaphor. Poetics Today 13/4: 574-606.
Gibbs, Raymond W. (1994). The Poetics of Mind. New York, Cambridge University Press.
Gineste, Marie-Dominique, Indurkhya, Bipin and Scart, Véronique. (2000). Emergence of
features in metaphor comprehension. Metaphor and Symbol, 15(3): 117-135.
Goodman, Nelson (1968). Languages of Art. An Approach to a Theory of Symbols. Indianapolis:
Bobbs-Merrill.
Indurkhya, Bipin (1986). Constrained Semantic Transference: A Formal Theory of Metaphor.
Synthese 68: 515-551.
Indurkhya, Bipin (1992). Metaphor and Cognition: an Interactionist Approach. Dordrecht:
Kluwer Academic Publishers.
Johnson, Mark (1987). The Body in the Mind: The Bodily Basis of meaning, Imagination, and
Reason. The Chicago: University of Chicago Press.
31
Keysar, Boaz and Glucksberg, Sam (1992). Metaphor and Communication. Poetics Today
13(4): 633-58.
Kittay, Eva F. (1987). Metaphor: Its Cognitive Force and Linguistic Structure. Oxford:
Clarendon.
Kövecses, Zoltán (2002).Metaphor. A Practical Introduction. Oxford: University Press.
Lakoff, George (1987). Women, Fire, and Dangerous Things: What Categories Reveal about
the Mind. Chicago: The University of Chicago Press.
Lakoff, George and Johnson, Mark (1980). Metaphors We Live By. Chicago: Univ. of Chicago
Press.
Lakoff, George and Turner, Mark (1989). More than Cool Reason: A Field Guide to Poetic
Metaphor. Chicago: The University of Chicago Press.
Leech, Geoffrey (1969). A Linguistic Guide to English Poetry. London: Longman.
Maalej, Zouhair (2001). Metaphor Making and Processing. Journal of Literary Semantics
XXX/3: 105-123.
MacCormac, Earl R. (1985). A Cognitive Theory of Metaphor. Cambridge: MIT Press.
Ortony, Andrew (1980). Some Psycholinguistic Aspects of Metaphor. In Cognition and
Figurative Language, Richard P. Honeck and Robert R. Hoffman (eds.), 69-87. New
Jersey: Lawrence Earlbaum Associates.
Ricoeur, Paul (1975). La Métaphore Vive. París: Editions du Seuil.
Recanati, François (2004). Literal Meaning. Cambridge: Cambridge University Press.
Romero, Esther (1990/1991). Las metáforas y el significado metafórico. La Balsa de la Medusa
15/16/17: 59-80.
Romero, Esther and Soria, Belén (1997/1998). Stylistic Analysis and Novel Metaphor.
Pragmalingüística 5/6: 373-89.
32
Romero, Esther and Soria, Belén (1998). Convention, Metaphor and Discourse. Atlantis 20.1:
145-159.
Romero, Esther and Soria, Belén (2003). ‘I’ as a Pure indexical in metonymic uses of
language. Talk presented in X Italo-Spanish Meeting of Analytic Philosophy:
Pamplona (Spain), November 6-8.
Romero, Esther and Soria, Belén (forthcoming(a)). Metaphoric Concepts and Language. In
Leonardi, P. (ed.), Padova Il Poligrafo.
Romero, Esther and Soria, Belén (forthcoming(b)). A View of Novel Metaphor in the Light
of Recanati’s Proposals. In Saying, Meaning and Refering. Essays on François
Recanati’s Philosophy of Language, María J. Frápolli (ed.) Palgrave Studies, in
Pragmatics, Language and Cognition.
Searle, John (1979). Metaphor. In Expression and Meaning, John Searle (ed.), 76-116.
Cambridge: Cambridge University Press. Reprinted in Metaphor and Thought, Andrew
Ortony (ed.), 92-123. Cambridge: Cambridge Univ. Press.
Sperber, Dan and Wilson, Deirdre (1986). Relevance: Communication and Cognition. Oxford:
Basil Blackwell.
Steen, Gerard (1994). Understanding Metaphor in Literature: An Empirical Approach. London
& New York: Longman.
Toolan, Michael (1996). Total Speech. An Integrational Linguistic Approach to Language.
Durham and London: Duke University Press.
Way, Eileen C. (1991). Knowledge Representation and Metaphor. Dordrecht: Kluwer
Academic Publishers.
Wilson, Deirdre and Sperber, Dan (2002). Truthfulness and Relevance. Mind 111: 583-632.
Department of Philosophy and Department of English Philology
33
University of Granada, Spain