1. modern linguistics as a change of paradigms

In metaphor, we are constrained not to violate the image-schematic structure of
the target; this entails that we are constrained not to violate whatever image-
schematic structure may be possessed by non-image components of the target.
(Turner 1990:252)
Lakoff calls these ‘target domain overrides’ (1993:216), and illustrates them with
give a kick and give an idea. When you give someone a kick, the person does not
‘have’ the kick afterward, and when you give someone an idea, you still ‘have’ the
idea. The target domain of transfer of energy or force does not allow that energy to
continue to exist after the transmission event, hence that metaphorical entailment
does not hold. Likewise, the target domain of knowledge does not imply that
knowledge transmitted is lost: that metaphorical entailment does not hold either.
The Invariance Hypothesis and the target domain override raise a fundamental
issue about conceptual metaphors: why do they exist in the first place? If the target
domain has image-schematic structure already, which can override the metaphor,
then why do we have metaphors? Likewise, if we can isolate image-schematic
structure, or construct highly schematic metaphors such as ORGANIZATION IS PHYSICAL STRUCTURE,isitnot simply a highly schematic conceptual structure that is instantiated in both source and target domains? If so, then is it really
a metaphorical mapping, or simply an instantiation of the image-schematic con-
ceptual structure in two different cognitive domains? (This latter view has been
propounded by Glucksberg [2001] and Jackendoff and Aaron [1991:328–30].)
Lakoff and Johnson present two counterarguments against these criticisms. First,
they argue that, although target domains of metaphors are structured, they are not
fully so: ‘they are not clearly enough delineated in their own terms to satisfy
the purposes of our day-to-day functioning’(Lakoff and Johnson 1980:118). Thus,
target domains lack at least some (image-schematic) structure. Lakoff and Johnson
argue that the linking of otherwise independent conceptual domains by metaphor
in fact creates similarity:
...the IDEAS ARE FOOD metaphor establishes similarities between ideas and
food. Both can be digested, swallowed, devoured, and warmed over, and both
can nourish you. These similarities do not exist independently of the metaphor.
The concept of swallowing food is independent of the metaphor, but the concept
of swallowing ideas arises only by virtue of the metaphor. (Lakoff and Johnson
1980:147–48)
The second counterargument that Lakoff and Johnson present for why concep-
tual metaphors exist is that there is an asymmetry between source domain and targetdomain. For example, love is expressed in terms of journeys, but journeys are notexpressed in terms of love (Lakoff and Johnson 1980:108). If image-schematicstructure were simply a highly schematic concept subsuming the correspondingconcepts in the source and target domains, then one would expect metaphoricalmappings to go in either direction; but they do not. Even when it appears thatthere is a bidirectional metaphorical mapping, Lakoff and Turner argue that thetwo mappings are different:
We can have cases like PEOPLE ARE MACHINES,asin


At the violet hour, when the eyes and back
Turn upward from the desk, when the human engine waits
Like a taxi throbbing, waiting (Eliot, The Waste Land)


and also the different metaphor MACHINES ARE PEOPLE,asinwhen we say,
‘The computer is punishing me by wiping out my buffer.’ But these are two
different metaphors, because the mappings go in opposite directions, and different
things get mapped.In MACHINES ARE PEOPLE, the will and desire of a person
are attributed to machines, but in the PEOPLE ARE MACHINES metaphor, there
is no mention of will and desire. What is mapped instead is that machines have
parts that function in certain ways, such as idling steadily or accelerating, that theymay break down and need to be fixed, and so on. (Lakoff and urner1989:132,
emphasis original)
These two counterarguments are persuasive; but they also imply that the Invariance
Hypothesis and the target domain override captures only part of the nature of
metaphorical mappings, and perhaps not the most important part. The fact that
both people and machines have parts that function (or malfunction) is part of why
human engine works as a metaphor. But Eliot is certainly conveying more than
that with the metaphor, including perhaps the mechanization of twentieth-century
life and its dehumanizing effects (not to mention further images created by the
simile on the next line).
It is likely that a far richer structure than simply compatible image-schemas is
brought into the target domain from the source domain. It also suggests that Lakoff
and Johnson’s first counterargument – the target domain lacks (image-schematic)
structure that is added by the metaphorical mapping from the source domain –
makes too sharp a distinction between target domain structure and mapped source
domain structure. It implies a minimum of interaction between the target domain
structure (already there) and the source domain structure (filling in for the absence of target domain structure). Instead, many metaphor theorists argue for amore interactive relationship between source and target domain structure, involving something like a ‘fusion’ or ‘superimposition’ of structure from both domains
(Jackendoff and Aaron 1991:334; they also cite Black’s [1979] ‘interaction’ and
Ricoeur’s [1978] ‘reverberation’). Jackendoff and Aaron suggest that the source
domain concepts are transformed as well in being metaphorically applied to the
target domain (ibid.). It is this intuition that blending theory attempts to capture
(see §8.3.3). This interactive relationship of course strengthens the first counter-
argument: the metaphor brings much more than extra image-schematic structure
to the target domain.
The final issue we wish to raise is the relationship between the conventionalmetaphors that Lakoff centers his attention on and novel metaphor creation. Lakoff and Johnson (1980:52–53) argue that some novel metaphors are extensions of existing conventional metaphors, such as the song lyric We’re driving in the fast lane on the freeway of love for LOVE IS A JOURNEY (Lakoff 1993:210). Lakoffand Johnson also allow for completely novel metaphors, using as an example LOVE IS A COLLABORATIVE WORK OF ART (Lakoff and Johnson 1980: 139–43).
They argue that a novel metaphor of this type can be a systematic mapping betweentwo conceptual domains.
Lakoff and Turner argue that other novel metaphors are more restricted; these
are image metaphors. An example they give is my wife . . . whose waist is an
hourglass ([3b] above; Lakoff and Turner 1989:90). Lakoff and Turner argue that inimage metaphors, specific and richly specified mental images are mapped, whereasin conventional image-schematic metaphors, ‘there is no rich imagistic detail’(Lakoff and Turner 1989:91). They also argue that image metaphors do not involvethe mapping of rich knowledge and inferential structure of conventional image-schematic metaphors (ibid.).
For conventional image-schematic metaphors themselves, Lakoff and Johnson
argue that they ultimately originate in human bodily and cultural experience. For
example, CONSCIOUS IS UP/UNCONSCIOUS IS DOWN, exemplified by wake
upand fall asleep, are based on the fact that ‘humans and most other mammals
sleep lying down and stand up when they awaken’ (Lakoff and Johnson 1980:15).
A cultural example is LABOR IS A (MATERIAL)RESOURCE,inthat it can be
measured, assigned a value, used and so on. This metaphor arises from the fact
that we use material resources for various purposes; by virtue of those purposes
the resources have value; and the use of those resources requires labor (Lakoff and
Johnson 1980: 65–66).
It seems likely that many conventional metaphors – the kind found in everyday
language – have a basis in everyday human experience. However, many novel
metaphors do not, and some conventional metaphors do not either, except in terms
of very general image schemas such as those described by Grady (1997, 1998).
Also, where Lakoff& Johnson discuss truly novel image-schematic metaphors
such as LOVE IS A COLLABORATIVE WORK OF ART, and Lakoff and Turnerdiscuss novel image metaphors, they do not describe how a speaker comes up with the new metaphor: they only describe what the structure of the metaphor is. But we have suggested that even conventional metaphor involves a blending of richer structure than just image-schematic structure between source and target domains.
It seems plausible that even conventional metaphors draw on the full richness of ourencyclopedic knowledge of our bodily and cultural experience, especially whenthey are first coined. Nor does there appear to be a difference in kind between the‘rich detail’ mapped in novel image metaphors and the ‘rich knowledge’ mappedin conventional image-schematic metaphors. If so, then there is only a differencein degree between conventional metaphors and novel metaphors.