1
Marga Reimer, University of Arizona, reimer@email.arizona.edu
Elisabeth Camp, Harvard Society of Fellows, ecamp@fas.harvard.edu
to appear in Handbook of Philosophy of Language, ed. E. Lepore and B. Smith (Oxford: Oxford
University Press, in preparation)
METAPHOR1
1. What is Metaphor?: A Tentative Characterization
Metaphor has traditionally been construed as a linguistic phenomenon: as something
produced and understood by speakers of natural language. So understood, metaphors are naturally
viewed as linguistic expressions of a particular type, or as linguistic expressions used in a
particular type of way. We adopt this linguistic conception of metaphor in what follows. In doing
so, we do not intend to rule out the possibility of non-linguistic forms of metaphor. Many
theorists think that non-linguistic objects (such as paintings or dance performances) or conceptual
structures (like love as a journey or argument as war2
) should also be treated as metaphors.
Indeed, the idea that metaphors are in the first instance conceptual phenomena, and linguistic
devices only derivatively, is the dominant view in what is now the dominant area of metaphor
research: cognitive science.3
In construing metaphor as linguistic, we merely intend to impose
appropriate constraints on a discussion whose focus is the understanding and analysis of
metaphor within contemporary philosophy of language.
Given this starting point, what can be said about metaphor that is not controversial? Very
little, as it turns out. Metaphor is a trope or figure of speech, where a ‘figure of speech’ is a nonliteral
use of language. This class also includes irony, metonymy, synecdoche, hyperbole, and
meiosis.4
What distinguishes metaphor from these other tropes? One standard definition of
1
We would like to thank Richard Moran, William Lycan, Emma Borg, Ram Neta, and Mike Harnish for
helpful comments on earlier drafts of this article.
2
For an extended discussion of these and other ‘conceptual metaphors,’ see Lakoff and Johnson (1980,
1999) and Lakoff (1993).
3
See the Center for the Cognitive Science of Metaphor Online
(http://philosophy.uoregon.edu/metaphor/metaphor.htm)
4
In irony, the intended meaning is in some sense the ‘contrary’ of the words uttered, as when one says of a
job that has clearly been poorly done, ‘Good job!’ In metonymy, a single characteristic or entity is used to
identify a more complex, related entity, as when ‘The White House’ is used to refer to the President.
Synecdoche is a kind of metonymy in which part of something is used to represent the whole, as in ‘All
hands on deck.’ Hyperbole involves exaggeration, meiosis understatement. When I say ‘These Tucson
summers are killing me,’ I am engaging in hyperbole; when I say of a wild party that things ‘got just a bit
out of hand,’ I am engaging in meiosis.
2
metaphor is as a figure of speech in which one thing is represented (or spoken of) as something
else. This construal of metaphor comports well with many examples of metaphor drawn from
classic literary works. Consider, for instance, “Juliet is the sun” (Shakespeare), “Time is the
devourer of all things” (Ovid), or “Poverty is the sister of beggary” (Aristophanes). In the first, a
girl, Juliet, is spoken of as the sun; in the second, time is spoken of as a ferocious beast; in the
third, poverty is spoken of as sister (and thus as a person).
Some philosophers, in an effort to explain metaphor’s characteristic rhetorical force, have
elaborated on this standard construal in terms of “representing-as.” Thus, Monroe Beardsley
(1967) identifies two features working in tandem within a metaphor. On the one hand, a metaphor
produces a conceptual tension between the concept that is expressed by the metaphorical term
and the concept(s) that we normally and intuitively apply to the subject. So, for example, there is
a ‘tension’ or mismatch between representing Juliet as the sun and as being a girl, or between
representing poverty as a sibling and as an economic state. Often (though, as we will see, not
always) this ‘tension’ renders the metaphorical sentence logically absurd if construed literally.
For this reason, Nelson Goodman (1968) characterizes the conceptual tension to which Beardsley
refers as involving a kind of “calculated category mistake.” A metaphor, he says, “projects” a set
of “labels” belonging to one realm of objects (e.g., celestial bodies) upon another realm to which
those labels do not ordinarily apply (e.g., human beings).
On the other hand, Beardsley points out, in spite of their absurdity metaphors are
generally quite intelligible and even profound. So, for example, Romeo’s metaphor seems to
serve as an effective means for communicating his feelings about Juliet (such as being dazzled by
her), to evoke similar attitudes in others, and to claim that she possesses certain properties (such
as being beautiful and attractive). Beardsley claims that metaphors are able to do this because the
sentence’s inherent conceptual tension imposes a “metaphorical twist” on the relevant term,
forcing it to refer to features with which it is normally merely associated.5
5
See Beardsley (1962).
3
These characterizations of metaphor do have a certain intuitive appeal, but they
themselves employ metaphorical language (“conceptual tension,” “label,” “projection”) in crucial
explanatory roles, and so fail to provide fully explicit and satisfactory theories of metaphor. As
we will see in what follows, this is quite typical. But it may also be unavoidable: as will also
become clear, metaphor is itself a vague and elusive phenomenon.
2. Metaphor and Contemporary Analytic Philosophy
Armed with this intuitive idea of what metaphor involves, let’s consider metaphor’s place
within analytic philosophy, broadly construed. We will then focus on how it has been treated
specifically by contemporary philosophers of language.
Although the last 30 years have seen an explosion of interest in metaphor within analytic
philosophy, the topic had previously been eschewed by analytic philosophers. Indeed, until Max
Black’s seminal (1962) paper “Metaphor,” it was virtually ignored. This was due largely to the
dominance of logical positivism during the preceding decades. Logical positivists viewed
metaphor as without cognitive significance, because they assumed that metaphors lacked the
crucial criterion for meaningfulness: verification conditions. Thus, consider T.S. Eliot’s line from
“The Love Song of J. Alfred Prufrock” (1917): “I have measured out my life with coffee spoons.”
It seems that nothing could possibly count as observational evidence for (or against) this claim,
because life, as an abstract entity, cannot in principle be measured in physical quantities. We
therefore have no idea what sort of situation, if observed, would demonstrate that the sentence
was true. From the fact that metaphors apparently fail to specify verification conditions, logical
positivists concluded that metaphorical speech lacks cognitive content altogether; instead, it must
merely serve to arouse feelings and images in its hearers.
Metaphor was thus mentioned by mid-century analytic philosophers only in order to be
set aside as irrelevant because unimportant to truth and knowledge. However, with the
publication of Black’s paper advocating an “interaction” theory of metaphor’s irreducible
“cognitive content,” analytic philosophers began to turn their attention to metaphor in earnest. In
4
the ’70’s and ’80’s thousands of scholarly papers on metaphor were published, along with many
anthologies devoted to metaphor. Several of the latter contained contributions not just by
philosophers of language, but also by literary theorists, philosophers of science, linguists,
psychologists, and cognitive scientists.6
The interest in metaphor among contemporary analytic
philosophers, and philosophers of language in particular, remains strong today. This is no doubt
due to a continued interest in natural, as opposed to formal or artificial, languages.
Philosophers of language have traditionally been interested in issues like the nature of
meaning and truth. And so, when they have turned their attention to metaphor, they have naturally
focused on these same issues. Before we turn to these particular topics, though, we should note
that virtually every area of analytic philosophy, broadly construed, has paid at least some serious
(if relatively limited) attention to metaphor.
Thus, within aesthetics, theorists have wanted to understand the special sort of ‘aptness’
and beauty that certain metaphors exhibit: the way in which a good metaphor can be, as Wallace
Stevens writes, “the cry of its occasion.”7
In philosophy of religion, there is interest in the
appropriate principles for interpreting religious texts, such as the Bible, metaphorically.8
Some
theologians and philosophers of religion believe that the nature of religious truth is such that it
can only be conveyed metaphorically.9
Epistemologists have considered the nature and utility of
analogical reasoning, which many cognitive psychologists believe to be crucially involved in the
interpretation of metaphor.10
Metaphysicians have been interested in the possibly metaphorical
status of crucial but theoretically troublesome terms, such as “existence” and “possible worlds.”11
Similarly, in philosophy of mathematics, there is talk of the metaphorical status of mathematical
concepts and truths.12
Finally, within philosophy of science, questions about the epistemic status
6
See, for instance, Sacks (1978), Ortony (1979), and Johnson (1981).
7
See Hills (1997); Hills cites the phrase from Stevens’ poem “An Ordinary Evening in New Haven.” See
also Isenberg (1973).
8
See Tracy (1979).
9
See Soskice (1987), McFague (1982).
10
See Gentner (1989) and Holyoak and Thagard (1995).
11
See Yablo (1996, 1998) and Walton (2000).
12
See Yablo (2000, 2002) and Lakoff and Nunez (1997).
5
of scientific models have been linked to the status of metaphors, which seem to bear important
structural similarities to models.13
3. Four Central Questions
Let us now turn our attention to the understanding and treatment of metaphor within
contemporary philosophy of language. Of the many questions concerning metaphor that have
been addressed within this particular area of philosophy, four stand out as especially central.
These are: (i) what are metaphors? (ii) what is metaphorical meaning? (iii) how do metaphors
work? and (iv) what is the nature of metaphorical truth? While these questions can be formulated
independently, they are logically connected insofar as the response given to any one constrains
possible responses to at least some of the others.
In addressing these questions, many philosophers have followed Black’s (1962)
methodological lead by first isolating a few uncontroversial cases of metaphor.14
These examples
in effect provide an extensional definition of “metaphor,” from which an explicit definition can
hopefully be derived. The benefit of this approach is that it gives us an intuitive, if vague, sense
for how metaphor differs both from literal language and from other figures of speech. The
drawback is that not all theorists begin with the same sorts of examples. Some focus on relatively
familiar, conversational metaphors like “You are the cream in my coffee,” or “I destroyed my
opponent’s argument,” while others attend to more novel, poetic metaphors such as “A
geometrical proof is a mousetrap,” or “Christ was a chronometer.” Employing such different
examples as paradigm cases raises the risk that the different parties will simply talk past one
another. The alternative, which is to provide a theoretical definition at the outset, is equally
problematic, simply because there are so few uncontroversial assumptions about metaphor.
We now spell out our four central questions in a bit more detail. In the next section, we’ll
see how various theories of metaphor have attempted to answer them.
13
See Hesse (1966, 1993), Kuhn (1979), and Boyd (1979).
14
This group includes both Davidson (1978) and Searle (1979).
6
(i) What are metaphors? Specifically, how does metaphorical language differ from literal
language and from other figures of speech? Philosophers have traditionally assumed that there is
an important in-principle difference between literal and figurative language, that figurative
language is essentially “marked” or distinctive, and that the figurative is in some sense a
“deviant” exploitation of the literal. These assumptions have recently come under scrutiny.
Thus, for instance, Sadock (1979), Rumelhart (1979), and Recanati (2001) have all questioned
whether there is a genuine difference in kind between literal and metaphorical language. Others,
such as Goodman (1968), Searle (1979), Sperber and Wilson (1986), and Nunberg (2002), have
rejected or at least downplayed the classical distinctions among different forms of figurative
language. Instead, they treat metaphor, simile, metonymy, synecdoche (and sometimes other
forms of figurative and non-literal language) as a single unified phenomenon.
(ii) What is the nature of metaphorical meaning? Answers to this question are
importantly tied to assumptions about what counts as “meaning” more generally. Thus, some
philosophers, like Grice (1975) and Searle (1979), argue that metaphors are like other forms of
indirect speech in expressing a distinctive speaker meaning which could in principle be stated
literally. Others, such as Black (1962) and Kittay (1987), argue that metaphors have a special,
irreducible sort of cognitive “meaning” or “significance,” but that this cannot be expressed in
literal language because it does not take the usual propositional form. Still others, such as
Davidson (1978) and Rorty (1987), agree that metaphors’ effects cannot be captured in
propositional form, but they conclude from this that metaphors have no meaning at all, on the
grounds that the only genuine candidates for “meaning” are truth-conditional, propositional
contents.
(iii) How do metaphors work? That is, how do metaphors manage to mean what they do?
This is perhaps the central ‘problem’ of metaphor, for the ease with which we are often able to
interpret metaphors, even subtle and complex ones, is rather puzzling on its face. In the case of
literal utterances, the interpretative process is presumably compositional. Oversimplifying a bit,
the hearer computes the utterance meaning on the basis of his or her grasp of individual word
7
meanings (where this includes fixing the values of any contextually-sensitive terms) and syntax.
Presumably something more is needed in the interpretation of metaphor, or else metaphorical
meaning would just be literal meaning. What ‘more’ could this be?
Some have thought that words themselves have special metaphorical meanings which
combine compositionally in the usual fashion, or that metaphorical meaning results from some
alteration in the process of composition itself.15
Others have assumed that metaphorical meaning
is computed by pragmatic means, employing global conversational principles, after the process of
literal composition is completed.16
Still others argue that metaphorical interpretation and
meaning have little to do with composition at all. They maintain instead that metaphors require
their hearers to engage in some special sort of cognitive process, such as cultivating an
“interplay” between two cognitive systems,17
or that they simply arouse, by merely causal
mechanisms, certain sorts of cognitive and affective effects in their hearers.18
(iv) What is the nature of metaphorical truth? Are metaphors associated with a distinctive
brand of truth? Here, the logical connection with the earlier questions is perhaps most obvious.
If the meaning of a metaphor is simply the proposition(s) the speaker intends to communicate,
and if these propositions can be given literal expression, then presumably literal and metaphorical
truth are identical in kind. If, on the other hand, metaphors are not in the business of
communicating propositions at all, but rather serve to evoke certain distinctive responses to
certain sorts of situations, then the relevant brand of truth, if any, must be quite different: perhaps
something more akin to “revealingness” or “comportment.”19
Finally, it might be the case that
metaphors do serve to communicate contents which can be true or false in the usual sense, but
that for one reason or another these contents are not capable of literal expression.
15
See Beardsley (1962), Levin (1977), Cohen and Margalit (1972), and more recently, Stern (2000).
16
See Grice (1975) and Searle (1979).
17
See Black (1962) and Kittay (1987).
18
See Davidson (1978) and Rorty (1987).
19
See Cooper (1993).
8
4. Four Influential Theories
Many theories of metaphor have been proposed and defended by philosophers of
language since the publication of Black’s “interaction” theory. Most attempt to answer questions
(i) through (iv), even if only indirectly. In this section, we survey four of the more influential
theories; this survey is intended to be representative rather than exhaustive.
(i) Simile Theories
Simile theories are the oldest and, until fairly recently, the most widely held theories of
metaphor.20
Aristotle seems to have been the first to suggest21
that metaphors are ‘compressed’ or
‘abbreviated’ similes. On any such theory, the meaning of a metaphor is identified with that of the
corresponding simile: where “A is B” is the metaphor (e.g., “Love is a journey”), its meaning is
given by the sentence “A is like B” (e.g., “Love is like a journey”). On such a view, the
interpretation of a metaphor is a matter of interpreting the corresponding simile, and the truth of
the metaphor is reduced to that of the simile.
The simile theory has both intuitive and methodological motivations. First, it often seems
as though some sort of comparison is made, or at least adumbrated, in metaphor. Consider
Hermann Melville’s (1856) “I had somehow slept off the fumes of vanity.” Although no
comparison is made explicitly here, it is nonetheless clear that Melville is drawing our attention to
similarities between the toxicity of fumes and the personality flaw of vanity. Second, the simile
theory appears to account for our conflicting intuitions about metaphors’ truth values. “Juliet is
the sun” is false if interpreted literally: Juliet is clearly not a gaseous ball of fire ninety-three
million miles from earth. But the simile that gives the sentence’s metaphorical meaning — “Juliet
is like the sun” — is arguably true.
20
We should also mention the so-called ‘substitution theory’ of metaphor, according to which a metaphor is
merely a substitute for some other expression which, used literally, would have expressed the same
‘cognitive content.’ This once-popular theory is no longer widely held, if it is held at all. For some
criticisms, see Black (1962).
21
We use ‘suggested’ here advisedly, as not all theorists agree that Aristotle actually endorsed this theory.
See Johnson (1981).
9
The most obvious methodological motivation for the theory lies in the reductive nature of
its central claim, that metaphor is a form of simile. At a minimum, the theory reduces two
problems to one: we now need only to explain how similes themselves work. Further, if the
meanings of similes are unproblematic because literal, then the ‘problem’ of metaphor has been
resolved altogether: we have analyzed metaphorical meaning and truth in terms of literal meaning
and truth.
Despite these virtues, the simile theory has been criticized on a number of counts.22
First,
not all metaphors are so readily translatable into simile form, if at all. William Lycan (1999)
makes the point nicely with the Shakespearean metaphor “When the blood burns, how prodigal
the soul/ Lends the tongue vows.”23
Concerning the simile that this metaphor is alleged to
abbreviate, Lycan (1999, p. 217) writes:
A first pass might be: When x, which is like a person’s blood, does something that
resembles burning, how prodigally y, which is like a person’s soul, does something
similar to lending some things that are vowlike to z, which resembles a person’s tongue.
He then remarks dryly, “We are not much wiser.”
Second, the simile analysis appears to represent metaphor as superficial and
uninformative. For many philosophers24
have claimed that similes themselves are trivial, on the
grounds that everything is like everything else in some respect or other. Yet metaphors often
appear to be informative and even profound. Third, and perhaps most importantly, the
similarities that we most naturally cite in explaining what a metaphor’s corresponding simile
means are often themselves figurative. Consider the opening lines of Sylvia Plath’s (1961) poem
“Mirror”: “I am silver and exact/I have no preconceptions.” Presumably the protagonist is here
describing herself metaphorically as a mirror; on the simile theory she thus means that she is like
a mirror. One natural elaboration of what this simile means is that she reflects the world around
22
See Black (1962), Beardsley (1967), Davidson (1978), Searle (1979), and Tirrell (1991).
23
Shakespeare, Hamlet, I.iii.116-7.
24
See Goodman (1968), Davidson (1978), and Searle (1979).
10
her, but the key word “reflects” here is itself obviously metaphorical. We seem to have fallen
into a vicious explanatory circle.
In response to the second and third of these worries, Robert Fogelin (1988) has proposed
a figurative version of the simile theory. According to Fogelin, statements of similarity should be
understood in terms of the notion of “salience”: the respect(s) in which A is said to be like B
depend on which of B’s features are salient in that context of utterance.25
What is distinctive
about figurative similes, and the metaphors that abbreviate them, is how those salient features are
determined. Consider the metaphor “Churchill was a bulldog.” According to Fogelin, in using
this metaphor we compare Churchill to a bulldog, but in order to understand this comparison we
must “trim the feature space” of bulldogs in terms of Churchill’s salient features.26
More
specifically: the hearer rules out a literal interpretation of the implicit simile on the grounds that
Churchill shares none of the usual, obviously salient features of bulldogs. The hearer nonetheless
charitably assumes that the alleged similarity does obtain, and so he ignores the salient features of
bulldogs that render the literal comparison false, such as having floppy ears and being small
enough to fit down burrows. He searches instead for features of bulldogs that match up with the
salient features of Churchill. Presumably, these include ‘character traits’ like resoluteness and
stubbornness; they might also include physical traits like having a thick neck and jowly face. The
metaphor “Churchill was a bulldog” claims that Churchill is like a bulldog in these respects.
Fogelin’s theory nicely defuses one of the main objections against the literal simile
theory: that it could not explain the informativeness and profundity of metaphor. While “Juliet is
like the sun” is literally false, according to Fogelin, it is true and even profound when interpreted
figuratively, because it raises to salience certain features that Juliet does share with the sun, and
that we might not otherwise notice. It also goes some way toward addressing the third worry, by
giving an analysis of figurative similes. However, it is doubtful that all figurative similes can in
fact be adequately analyzed in terms of features which the two objects in question are believed to
25
In this Fogelin follows Tversky (1977).
26
Fogelin (1988), p. 91.
11
actually share, as Fogelin assumes.27
So, for instance, “Sally is a block of ice” is intuitively true
just in case Sally is like ice in being cold. But there’s no obvious single property, of coldness,
which applies to both frozen water and personal temperaments in the same way, and it’s not clear
how to analyze the simile further into features that are in fact shared. Similarly, the sense in
which Juliet is like the sun intuitively depends in part on a higher-order analogy between the
properties of being bright and being beautiful, rather than upon a concrete feature possessed by
both Juliet and the sun.28
Finally, Fogelin’s view is clearly still vulnerable to the first objection
above, that not all metaphors can be translated into simile form.
(ii) Interaction Theories
As we mentioned, one of the earliest modern alternatives to the simile theory was the
‘interaction’ view of metaphor. This view was first advocated by the literary theorist I.A.
Richards (1936), and was subsequently developed by the philosopher Max Black (1962). Such
theories have two central claims: (i) that metaphors have an irreducible “cognitive content,” and
(ii) that this cognitive content (or “meaning”) is produced by the “interaction” of different
cognitive systems. Interactionists generally claim that the “cognitive contents” of metaphors can
be true, even though they are not amenable to literal expression.
According to Black, in a metaphor of the form “A is B,” the “system of associated
commonplaces” for B “interacts” with that of A to generate a metaphorical meaning for the whole
sentence, by “filtering” our thoughts about the ‘system’ associated with A. Consider one of
Black’s examples: “Man is a wolf.” The properties of being a predator, traveling in packs, and
being fierce and ruthless are all commonplaces associated with “wolf.” These properties are
therefore instrumental to comprehending the metaphor: they serve as the “filter” for thinking
about mankind, by emphasizing just those commonplaces of “man” that fit with them. The
27
See Ortony (1979), Searle (1979).
28
In this case, we can construct a higher-order property which Juliet and the sun do share: the property of
possessing a property which bears a certain relation R to other properties. But then we seem to be back in
the situation of postulating uninformative analyses of the sort criticized by Lycan above.
12
metaphor’s “cognitive content” or meaning is the distinctive way of thinking about mankind that
this filtering produces. Notice here that “commonplaces” need not be true. For instance, the
commonplace that wolves are ruthless is still part of the relevant system, even though wolves, as
amoral creatures, cannot actually be ruthless. Likewise, the other commonplaces mentioned
above would remain relevant even if it turned out that in fact wolves are docile herbivores who
tend to travel in pairs. What matters is not the actual properties of the objects denoted, or even
the properties that speakers and hearers believe those objects to possess, but rather what the
denoting expressions “call to mind.”29
The interaction theory’s central motivation is to account for the apparent fact that
metaphors function as powerful cognitive tools: as devices that enable us to better understand the
world in which we live. It thus coheres nicely with the view, advocated by Thomas Kuhn (1979)
and Richard Boyd (1979) among others, that scientific models appear to increase scientists’
understanding of the universe. The interaction theory also comports well with the view, popular
among certain cognitive scientists, that ordinary thought and reason are largely, and irreducibly,
metaphorical.30
For the interactionist regards as misguided any attempt to reduce metaphorical
meaning to literal meaning.
Perhaps not surprisingly, though, the interaction theory as presented by Black has seemed
too vague to be of great theoretical value. Part of the problem is, once again, that Black analyzes
metaphor itself in terms of other metaphors like “association,” “interaction,” and “filtering.”
Nevertheless, some theorists have managed to develop Black’s central claim — that metaphor is
irreducibly a matter of interacting cognitive systems — in more theoretically tractable terms.31
Rather different criticisms have been launched by Donald Davidson (1978) and Fogelin
(1988).32
Davidson claims that there is no clear theoretical value to positing special metaphorical
29
“Commonplaces” can still call features to mind even if they are not believed to be true of the objects
denoted by the relevant term. For instance, even if both the speaker and hearer know that gorillas are in
fact gentle creatures, the stereotype that gorillas are nasty and violent can play a role in determining the
“cognitive content” of the metaphorical utterance.
30
See Lakoff and Johnson (1980, 1999) and Lakoff (1993).
31
See e.g., Lakoff and Johnson (1980), and Kittay (1987).
32
For Black’s reply to Davidson, see his (1978).
13
‘meanings’ or ‘cognitive contents’.33
As he puts it, to say that metaphorical meaning explains
how metaphor works is “like explaining why a pill puts you to sleep by saying it has a dormative
power”34
: we have simply found a new, fancier way to describe the phenomenon under
investigation, but we have made no real explanatory progress. A second objection of Davidson’s
concerns Black’s claim that metaphors are not amenable to precise literal paraphrase. If so, asks
Davidson, why should we suppose that there is any meaning there to begin with? If metaphors
have a “cognitive content” beyond the literal, then why should it be so difficult, even impossible,
to capture that content in literal language? Finally, Fogelin points out that not all cases of
metaphor are so easily explained in terms of conceptual “interaction.” Consider John Keats’s
(1819) metaphor: “O for a beaker fill of the warm south.” One would be hard pressed to specify
the “cognitive systems” whose “interaction” makes this such an effective use of language. More
generally, Black’s view works best for metaphors that consist of a general kind term predicated of
an individual or kind, but not all metaphors take this form.35
Philosophers suspicious of the special, irreducible “cognitive contents” posited by
interactionists have developed several alternatives, the best known of which are Gricean and
noncognitivist theories. Let us consider these in turn.
(iii) Gricean Theories
Gricean theories of metaphor are in the first instance theories of metaphorical
interpretation. Their central claim is that to understand a metaphor is to understand what a
speaker intends to communicate by means of it, where communication is analyzed in Gricean
terms. Roughly, successful communication consists in the hearer’s recognizing the speaker’s
intention to get the hearer to recognize what she is trying to communicate to him. Insofar as a
metaphor can be said to have a meaning, this is identified with what the speaker intends to
33
See Kittay (1987) for a response to Davidson’s contention here.
34
Davidson (1978), p. 31.
35
See White (1996).
14
communicate; the sentence uttered itself has only its literal meaning. A metaphor’s truth-value is
reduced to that of what the speaker intends to communicate.
Since John Searle is the best-known advocate of a broadly Gricean theory of metaphor,36
we will consider his view. According to Searle (1979, pp. 76-8):
The problem of explaining how metaphors work is a special case of the general problem
of explaining how speaker meaning and sentence or word meaning come apart... Our task
in constructing a theory of metaphor is to try to state the principles which relate literal
sentence meaning to metaphorical [speaker’s] utterance meaning.
Searle divides the interpretative process into three stages. First, the hearer must decide whether to
look for a nonliteral interpretation. Such a search is typically undertaken when a literal
interpretation of the utterance would yield a defective (false, trivial, or irrelevant) utterance.
Second, once the hearer decides to seek a specifically metaphorical interpretation, she employs a
set of principles for generating possible speaker meanings. Searle offers eight principles
according to which the uttered phrase can “call to mind” a different meaning “in ways that are
specific to metaphor.”37
So, for instance, the intended meaning might specify a defining feature
or a well-known characteristic associated with the uttered phrase. Third, the hearer must identify
which among this range of possible meanings is most likely to be the speaker’s intended meaning.
Thus, the fact that islands are great places to vacation is not likely to be relevant to the
interpretation of Donne’s metaphor “No man is an island,” while the fact that islands are
separated from other land masses is likely to be relevant, especially in the context of a speech
about social relations.
The motivation for a Gricean account is three-fold. First, it captures the intuition that
metaphors are meaningful, that they have a “cognitive content” other than their literal content.
Second, it does this without violating what Grice (1975) called “Modified Occam’s Razor.” This
methodological principle is simply Occam’s Razor applied to linguistic meanings: Don’t multiply
36
See Grice (1975) for a brief sketch of such a view; see also Martinich (1984).
37
Searle (1979), p. 85.
15
senses beyond necessity. The Gricean account respects this principle because it explains
metaphors’ meanings by appealing only to literal sentence meaning plus general interpretive
principles. And third, a Gricean theory embeds the explanation of metaphor within a welldeveloped,
independently motivated theory of linguistic communication that accommodates a
variety of cases where sentence meaning and speaker meaning appear to come apart.
Criticisms of Gricean theories of metaphor are varied. First, on most Gricean accounts,
the hearer must first identify the utterance as somehow “defective” if interpreted literally. Only
then is the search for an alternative, non-literal interpretation triggered.38
However, not all
utterances used metaphorically are obviously defective.39
A sentence like “No man is an island”
exhibits no grammatical deviance: it is literally true, albeit trivially so. Utterances of sentences
like “The rock is becoming brittle with age” or “Anchorage is a cold city” could plausibly count
as both true and informative when construed either literally or metaphorically, depending on the
context. And a “twice true” metaphor like “Jesus was a carpenter” could count as both literally
and metaphorically true and informative within a single context of utterance. It seems that a
speaker could even plausibly intend to communicate both contents simultaneously. So there need
be no deviance either in the sentence itself or in the utterance of it.
Second, there is empirical evidence to suggest that the literal meaning of a sentence used
metaphorically needn’t actually be processed in order for the metaphor to be understood.40
If this
is correct, then it seems that a Gricean theory could at most serve as a rational reconstruction,
rather than a factual description, of the interpretive process.41
Third, on a Gricean theory, the
speaker’s communicative intentions exhaust the metaphor’s meaning. Yet a metaphor’s import
often seems to go beyond what the speaker explicitly anticipated, especially for novel, poetic
metaphors.42
Finally, Griceans generally assume that metaphorical meaning, like speaker
38
Searle allows that a metaphorical interpretation may be triggered without any defectiveness, for instance
when we are on the lookout for them while reading a Romantic poem (p. 105).
39
See Reddy (1969), Cohen (1975), and Tirrell (1991).
40
See Rumelhart (1979) and Gibbs (1994).
41
This seems to be how Searle, at least, intends his view to be understood.
42
Though the speaker’s explicit expectations may not exhaust, or even track, the content of her actual
intentions. See, e.g., Cavell (1976).
16
meaning more generally, is fully propositional in form and fully capable of literal expression.
The Gricean theory thus seems doomed to leave out what is most interesting about metaphor: its
complex cognitive and affective “import,” which seems to be inherently inexpressible in literal
terms.43
(iv) Noncognitivist Theories
In light of the difficulties we’ve encountered so far, some contemporary philosophers of
language have questioned the widely-held view that metaphors are, in any substantive sense,
meaningful. These philosophers — “noncognitivists” — do not question metaphor’s effectiveness,
only the means by which its effects are achieved. The central claim of such theorists is that a
sentence used metaphorically has no cognitive content aside from its literal content.
Noncognitivists thus resemble Griceans in denying that the words uttered themselves have any
special meaning. They depart from Griceans, though, in also denying that there is any determinate
propositional thought which the speaker intends to communicate by means of those words. These
negative claims are often coupled with positive views about how metaphor does “work its
wonders” after all. Thus, Davidson (1978) offers what might be termed a “causal theory” of
metaphor.44
On his view, “a metaphor makes us attend to some likeness, often a novel or
surprising likeness, between two or more things,” by making us “see one thing as another.”45
Despite its undeniable counter-intuitiveness, noncognitivism is not without motivation.
First, it accounts for the facts that many metaphors don’t easily admit of literal paraphrase, and
that their ‘import’ seems to be different in kind from that of typical literal utterances. Second, it is
remarkably economical: it purports to explain how metaphor works without appealing to special
word meanings or even to Gricean speaker meanings. According to Davidson, a metaphor is like
a bump on the head, or a drug: one can employ it to cause certain effects in one’s audience,
43
Though see Camp (2003) for defense of a broadly Gricean theory on which at least some metaphorical
utterances can be intended to communicate complex structured representations which are not fully
propositional and which may not be fully explicitly appreciated by the speaker.
44
See Rorty (1987) for further discussion of the merely causal status of metaphor on this view.
45
Davidson (1978), pp. 31, 45.
17
including noticing surprising similarities between objects, but this should not lead us to suppose
that the metaphor, the bump or the drug itself means those effects, or even that the agent meant
that effect by her action. Third, Davidson argues, the analogy with similes actually supports
noncognitivism. We are much less tempted to suppose that similes have a special meaning
beyond their literal meaning: “Juliet is like the sun” means that Juliet is like the sun, nothing
more, nothing less.46
Of course, the point of uttering the simile would not be merely to express
that proposition, but rather to draw the hearer’s attention to similarities between Juliet and the
sun. But we needn’t then suppose that the speaker means to claim that those similarities are there
to be noticed.
The noncognitivist theory has been criticized on a variety of grounds.47
Most obviously,
the theory seems to conflict with fact that metaphors are cognitively significant: that they can be
understood or misunderstood, that they figure in our reasoning and thought, and that they can be
true or false. Moreover, as Merrie Bergmann (1982) and others have pointed out, a
noncognitivist view misses the role that metaphors play in assertion and counter-assertion. If I
call Bill a vulture, and you deny this, then it seems clear that something has been asserted and
denied, and that this ‘something’ is not the claim that Bill is a certain kind of bird.48
Finally, as several philosophers49
have pointed out, the noncognitivist view appears to be
incompatible with the phenomenon of dead metaphors. Dead metaphors are expressions which
have lost their metaphorical import through frequent use, and so no longer invite creative
interpretation. Their former metaphorical import has ‘hardened’ into a new literal meaning. Thus,
the expression “burned up,” as in “He was all burned up about his impending divorce,” is a dead
metaphor, whose second literal meaning is just extremely angry. As Davidson puts it, the
expression no longer conjures up “fire in the eyes or smoke coming out of the ears.”50
The
difficulty this poses for the noncognitivist is that it seems as if dead metaphors could only acquire
46
Here, there is explicit disagreement with Fogelin (1988).
47
See Moran (1989 and 1996) for these and other objections.
48
See Bezuidenhout (2001) for further discussion of this point.
49
See Goodman (1978), Moran (1989), and Reimer (1996).
50
Davidson (1978), p. 36.
18
their secondary literal meanings if they were previously used to communicate those very
meanings. And this would seem conflict with the basic noncognitivist commitment: that speakers
do not mean anything by metaphors.51
5. Current and Future Trends
The most active domain of metaphor research is currently located in cognitive science.52
The focus here is typically on issues such as metaphor’s influence on thought and action, and the
role of metaphor in cognitive development and linguistic competence.53
There is generally less
emphasis on metaphor as a form of expression in natural language. Thus, for instance, George
Lakoff and his colleagues are most interested in metaphor as a cognitive tool for extending
concepts’ initial applications to new realms. They argue that we metaphorically transfer basic
physical concepts like up and over to other realms: to the social, emotional, scientific, and even
mathematical domains. These metaphorical mappings render certain ways of speaking and acting
natural (e.g., “He’s moving up in the world,” “I’m feeling quite up today”). What we would
normally classify as metaphorical language should, on this view, be analyzed instead as a direct,
explicit representation of metaphorical way of thinking.54
The future of metaphor research within the philosophy of language itself is less clear.
One hope is that philosophers of language will work with, or at least alongside, researchers in
other disciplines, so that their theories can be informed and even shaped by the varied
observations garnered from these other disciplines. In particular, many of the theories discussed
above invoke the notions of ‘salience’ and ‘similarity’ in one way or another, but have little to say
about what these involve. Researchers in linguistics, cognitive science, psychology, and
neurobiology are developing such notions, in work on metaphor and on other areas.
To see how philosophical theories of metaphor could be informed by other disciplines,
51
But see Reimer (2001) for a defense of Davidson.
52
See Gibbs (1994).
53
See, for instance, Happé (1995) and Langdon, Davies, and Coltheart (2001).
54
See Lakoff (1993), Lakoff and Turner (1989).
19
we need only return to the four questions we discussed above as central concerns for philosophers
of language:
(i) What is metaphor?
(ii) What is the nature of metaphorical meaning?
(iii) What is the nature of metaphorical communication?
(iv) What is the nature of metaphorical truth?
Interest in these questions is by no means confined to philosophy of language. Literary theorists
are interested in distinguishing metaphor from other figures, such as simile and irony, and thereby
address (i). Cognitive scientists do so as well, by proposing that metaphors be viewed primarily
as mental representations and only derivatively as linguistic phenomena. Philosophers and
historians of science argue, along with cognitive scientists, that metaphors are significant
cognitive tools, and in this way they address (ii). Cognitive psychologists and linguists have done
significant empirical research on the processing of metaphor and of language generally, thus
shedding light on (iii). Interest in (iv) is perhaps more fully limited to philosophy per se, but
metaphysicians, philosophers of mathematics, and even some historians of science have been
concerned to explore the possibility of a substantive sort of truth which is not literal. It should
thus be clear that philosophers of language can learn much about metaphor from the research
efforts of those outside of their own area, and perhaps even especially outside of philosophy
itself.
At the same time, an increasing focus on various forms of context-sensitivity within
philosophy of language and linguistics has led to the development of new explanatory tools and
strategies. “Hidden indexicals,”55
“unarticulated constituents,”56
and “free enrichment”57
have all
been postulated as mechanisms for bridging the gap between what might seem to be a sentence’s
semantically encoded content and the content that is expressed by an utterance of it on a given
55
Cf. Stanley (2000)
56
Cf. Perry (1986) and Crimmins (1992).
57
Cf. Recanati (1995).
20
occasion. These same mechanisms have recently begun to be deployed in explaining metaphor as
well. For instance, Josef Stern has argued that metaphors function like demonstrative terms.58
David Hills59
and Kendall Walton60
have each argued that metaphorical meaning crucially
depends upon a sort of “pretense” or “make-believe” about word use, and about the schema of
objects invoked. And Anne Bezuidenhout61
and François Recanati62
have advocated treating
metaphor as a form of “direct expression” rather than of implicature or indirect speech, as Grice
and Searle do. Perhaps this influx of new ideas will lead to a fruitful reconfiguration of the
established options, much as occurred around the time of Davidson’s and Searle’s seminal
publications in the late 1970’s.
58
Stern (2000).
59
Hills (1997).
60
Walton (1993)
61
Bezuidenhout (2001).
62
Recanati (2001).
21
References
Beardsley, M. (1967): “Metaphor,” in The Encyclopedia of Philosophy, vol. 5, ed. P. Edwards
(New York: Macmillan).
Beardsley, M. (1962): “The Metaphorical Twist,” Philosophy and Phenomenological Research 22:3.
Bergmann, M. (1982): “Metaphorical Assertions,” Philosophical Review 91, 225-245.
Bezuidenhout, A. (2001): “Metaphor and What is Said: A Defense of a Direct Expression View of
Metaphor,” Midwest Studies in Philosophy 25, 156-186.
Black, M. (1962): “Metaphor,” in Models and Metaphors (Ithaca: Cornell University Press).
Black, M. (1978): “How Metaphors Work: A Reply to Donald Davidson” in On Metaphor, ed. S. Sacks
(Chicago: University of Chicago Press).
Boyd, R. (1979): “Metaphor and Theory Change: What is ‘Metaphor’ a Metaphor For?” in
Metaphor and Thought, ed. A. Ortony (Cambridge: Cambridge University Press).
Camp, E. (2003): Saying and Seeing-as: The Linguistic Uses and Cognitive Effects of Metaphor,
dissertation, University of California, Berkeley.
Cavell, S. (1976): “A Matter of Meaning It,” in S. Cavell, Must We Mean What We Say?
(Cambridge: Cambridge University Press).
Cohen, L. J. and A. Margalit (1972): “The Role of Inductive Reasoning in the Interpretation of
Metaphor,” in Semantics of Natural Language, ed. D. Davidson and G. Harman (Dordrecht:
D. Reidel).
Cohen, T. (1975): “Figurative Speech and Figurative Acts,” Journal of Philosophy 71, 669-684.
Cooper, David (1993): “Truth and Metaphor” in Metaphor and Knowledge: Knowledge and
Language, vol. III, ed. F.R. Ankersmit and J.J.A. Mooij (Dordrecht: Kluwer Academic
Publishers).
Crimmins, M. (1992): Talk about Beliefs (Cambridge, MA: MIT Press).
Davidson, D. (1978): “What Metaphors Mean,” in On Metaphor, ed. S. Sacks. (Chicago:
University of Chicago Press).
Eliot, T. S. (1917), “The Love Song of J. Alfred Prufrock,” reprinted in Collected Poems, 1909-1962
(London: Faber and Faber, 1974).
Fogelin, R. (1988): Figuratively Speaking (New Haven: Yale).
Gentner, D (1989): “The Mechanisms of Analogical Learning,” in Similarity and Analogical Reasoning,
ed. A. Ortony and S. Vosniadou (Cambridge: Cambridge University Press).
Gibbs, R. (1994): The Poetics of Mind (Cambridge: Cambridge University Press).
Goodman, N. (1968): Languages of Art: An Approach to a Theory of Symbols (Indianapolis:
Hackett).
Goodman, N. (1978): “Metaphor as Moonlighting,” in On Metaphor, ed. S. Sacks. (Chicago:
University of Chicago Press).
Grice, H.P. (1975): “Logic and Conversation,” Speech Acts: Syntax and Semantics vol. 3, ed. P.
Cole and J. L. Morgan. (New York: Academic Press).
Happé, F. G. E. (1995): “Understanding Minds and Metaphors: Insights from the Study of
Figurative Language in Autism,” Metaphor and Symbolic Activity 10, 275-95.
Hesse, M. (1966): “The Explanatory Function of Metaphors,” in M. Hesse, Models and Analogies
in Science (Notre Dame, Quebec: Notre Dame University Press), 157-177.
Hesse, M. (1993): “Models, Metaphors and Truth,” in Metaphor and Knowledge: Knowledge
and Language, vol. III, ed. F.R. Ankersmit and J.J.A. Mooij (Dordrecht: Kluwer Academic
Publishers).
Hills, D. (1997): “Aptness and Truth in Verbal Metaphor,” Philosophical Topics 25:1, 117-153.
Holyoak, K. and P. Thagard (1995): Mental Leaps: Analogy in Creative Thought (Cambridge, MA: MIT
Press).
Isenberg, A. (1973): “On Defining Metaphor,” in Aesthetics and Theory of Criticism: Selected
Essays of Arnold Isenberg, ed. Callagan, Cauman, and Hempel (Chicago: University of
22
Chicago Press).
Johnson, M. (1981) ed.: Philosophical Perspectives on Metaphor (Minneapolis: University of
Minneapolis Press).
Kittay, E. (1987): Metaphor: Its Cognitive Force and Linguistic Structure (Oxford: Oxford
University Press).
Kuhn, T. (1979): “Metaphor in Science,” in Metaphor and Thought, ed. A. Ortony (Cambridge:
Cambridge University Press).
Lakoff, G. (1993): “The Contemporary Theory of Metaphor,” in Metaphor and Thought. 2nd
ed.,
ed. A. Ortony, (Cambridge: Cambridge University Press), 202-251.
Lakoff, G. and M. Johnson (1980): Metaphors We Live By (Chicago: University of Chicago
Press).
Lakoff, G. and R. Nunez (1997): “The Metaphorical Structure of Mathematics: Sketching Out
Cognitive Foundations for a Mind-Based Mathematics,” in Mathematical Reasoning:
Analogies, Metaphors, and Images, ed. L. English (Hillsdale, N. J.: Erlbaum).
Lakoff G. and M. Turner (1989): More than Cool Reason: A Field Guide to Poetic Metaphor
(Chicago: University of Chicago Press).
Langdon, R., M. Davies and M. Coltheart (2002): “Understanding Minds and Understanding
Communicated Meanings in Schizophrenia,” Mind and Language 17:1/2, 68-104.
Levin, S. (1977): The Semantics of Metaphor (Baltimore: Johns Hopkins University Press).
Lycan, W. (1999): Philosophy of Language: A Contemporary Introduction (London and New
York: Routledge).
Martinich, A.P. (1984): “A Theory for Metaphor,” Journal of Literary Semantics 13, 35-56.
McFague, S. (1982): Metaphorical Theology: Models Of God In Religious Language
(Philadelphia: Fortress Press).
Moran, R. (1989): “Seeing and Believing: Metaphor, Image, and Force,” Critical Inquiry 16, 87-
112.
Moran, R. (1997): “Metaphor,” in A Companion to Philosophy of Language ed. C. Wright and R.
Hale (Oxford: Basil Blackwell).
Nunberg, G. (2002): “The Pragmatics of Deferred Interpretation,” in The Blackwell Encyclopedia
of Pragmatics, ed. Laurence Horn and Gregory Ward (Oxford: Blackwell Publishers).
Ortony, Andrew (1979): “Beyond Literal Similarity,” Psychological Review 87, 161-80.
Ortony, A. (1979) ed.: Metaphor and Thought (Cambridge: Cambridge University Press); 2nd
edition (1993).
Perry J. (1986)” “Thought without Representation,” Proceedings of the Aristotelian Society 60:
137-151.
Recanati, F. (2001): “Literal/Nonliteral,” Midwest Studies in Philosophy 25, 264-274.
——— (1995): “The Alleged Priority of Literal Interpretation.” Cognitive Science 19, 207-232.
Reddy, M. J. (1969): “A Semantic Approach to Metaphor,” in Papers from the Fifth Regional
Meeting of the Chicago Linguistics Society, ed. R. I. Binnick et al (Department of Linguistics,
University of Chicago).
Reimer, M. (2001): “Davidson on Metaphor,” Midwest Studies in Philosophy 25, 142-155.
Reimer, M. (1996) “The Problem of Dead Metaphors,” Philosophical Studies 82:1, 13-25.
Richards, I. A. (1936): The Philosophy of Rhetoric (Oxford: Oxford University Press).
Rorty, R. (1987): “Unfamiliar Noises I: Hesse and Davidson on Metaphor,” Proceedings of the
Aristotelian Society (supp. volume 61), 283-296.
Sacks, S. (1978): On Metaphor (Chicago: University of Chicago Press).
Searle, J. (1979): “Metaphor,” in Expression and Meaning: Studies in the Theory of Speech Acts
(Cambridge: Cambridge University Press).
Saddock, J. (1979): “Figurative Speech and Linguistics,” in Metaphor and Thought, ed. A.
Ortony (Cambridge: Cambridge University Press).
Soskice, J. M. (1987): Metaphor and Religious Language (Oxford: Clarendon Press).
23
Sperber, D. and D. Wilson (1986): “Loose Talk,” Proceedings of the Aristotelian Society 86, 153-
171.
Stanley, J. (2000): “Context and Logical Form,” Linguistics and Philosophy 23, 391–434.
Stern, J. (2000): Metaphor in Context (Cambridge, MA: MIT Press).
Tirrell, L. (1991): “Reductive and Nonreductive Simile Theories of Metaphor,” Journal of
Philosophy 88:7, 337-358.
Tversky, A. (1977): “Features of Similarity,” Psychological Review 84:4, 327-352.
Walton, K. (1993): “Metaphor and Prop-oriented Make-Believe,” European Journal of
Philosophy 1, 39-57.
White, Roger (1996): The Structure of Metaphor: The Way the Language of Metaphor Works
(Oxford: Blackwell).