Systematicity in Metaphor and the Use of Invariant Mappings A M Wallin gton School of Computer Science, University of Birmingham, Birmingham, B152TT , UK[email protected]AbstractI argue that the origin of the apparent systematicity found in families of related metaphors lies not in ontologically rich skeletal source domain schemas with slots that map to target domain correspondents. This suggests a rigidity I argue is inappropriate and misses significant cross metaphor generalizations. Instead, I will claim that metaphors utilise just a few core source-target correspondences. Users can extend and elaborate upon these and by doing so, give the impression of a systematic exploitation of a domain, by incorporating into the utterance any aspect of the user’s encyclopaedic knowledge that can be linked to the core correspondences. However, this linkage is subject to the constraint that any conclusions then drawn about the source domain meaning must ultimately be gro unded in spe cifi c types of information tha t transfer inv aria ntl y bet wee n source and target as adjuncts to the core correspondences. It is in these invariant mappings that systematicity is to be found. 1 Introduction In this paper, I will attempt to give an account of why metaphorical utterances appear to abide by a principle of systematicity, that is a tendency to cluster into families of metaphors sharing common source and target domains. For example, metaphorical utterances pertaining to love are often viewed or described as being about a journey , an argument is viewed as war, theories are viewed as buildings and s o on. One explanation for this systematicity is that the metaphorfami lies consi st of a fixed and stable set of map pin gs or corresp ond ences between the linked source and target domains. For example, LOVE AS A JOURNEY assumes correspondences between lovers and travellers, the love relationship and a vehicle, problems in the relationship and obstacles in the path, and so on. 1 Expressi ons such as “W e ar e at a crossro ads in ou r relationship”, and “We may have to go our separate ways” are consistent with
39
Embed
Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
these correspondences. These related source-target correspondences would be
termed, if we were to adopt a cognitive linguistic approach, “conceptual
metaphors” and many hold them to be “a fixed part of our conceptual system.”
(Lakoff 1993:208).
However, I will argue that we need to reject such a static approach, in which the
correspondences involved in the individual conceptual metaphors are a fixed
part of our conceptual system. I believe there are problems both with the
assumption of a fixed and stable set of detailed correspondences and with the
proper definition of a domain and these will be discussed in sections two and
three. The main claim however of the paper is that significant generalisations
concerning what is common to metaphor are being missed.
Instead, I will argue that the production and interpretation of metaphorical
utterances must be seen as part of dynamic process that takes into account
metaphor producers’ (encyclopaedic) knowledge, the knowledge they assume
their co-locutors to have, the linguistic, social and cultural context of the
discussion and the situation in which the conversation takes place. Along with
recent work on Relevance Theory and metaphor (see Vega Moreno 2007,
Carston 2002, Sperber and Wilson 2008; see also Hobbs 1990 for similar
proposals and the work of Glucksberg and associates e.g., Glucksberg 2001) I
assume that metaphor interpretation involves (often extensive) inferencing, or
searching through knowledge associated with the source words for specific
types of information that can answer queries thrown up by the surroundingdiscourse.2 Metaphor interpretation is highly context dependent, as also noted
by Stern, 2000; Steen, 2007, Steen et al., this volume; Brouwer, this volume).
But if this is the case, then what accounts for the apparent systematicity? The
Relevance-Theoretic accounts do not give an account of systematicity and
Glucksberg and associates argue against it (e.g. McGlone, 2007). To attempt an
answer I shall first propose a metaphor for the process of metaphor use: that of a
jazz musician improvising around a theme. Importantly, there are principles and
constraints governing the process of improvisation and what I shall attempt to
do in the first part of this paper is to suggest general means, not tied to specific
conceptual metaphors, by which a very basic, central, or core mapping between
a source and target (of a similar nature to Grady’s “Primary Metaphors”, (Grady,
1997a,b)) can be extended and elaborated upon in order to allow specific types
of information to be inferred. This ability to extend and elaborate a core
mapping between a source and target -improvise around a theme- will give the
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
types of invariant information that can accompany such mappings, the body of
knowledge that allows the relevant inferences to be made and an expectation
that the reader will make necessary inferences and utilise the core mapping.
Given these assumptions, many different metaphorical utterances will, as with
standard Conceptual Metaphor Theory, appear to be systematically related toeach other. Different utterances may exploit different inference chains in order
to convey different types of information, but since the invariant mappings
conveying the information must accompany, or act as an adjunct to, a core
source-target mapping, the entities referred to in the different utterances must all
relate to the source in the core source-target mapping. Furthermore, since the
producer of the metaphor must assume that the listener will easily draw the
correct inferences if communication is to succeed -at least for mundane, non-
poetic, discourse-, then the different entities, attributes, relations and so on that
are used to extend or elaborate on the core mapping are likely to be quite closely
and conventionally related to the source part of the core mapping.
If they are quite closely related, then they might loosely and or conveniently be
described as belonging to the same domain, but by focussing on the
identification of domains we ignore the significant generalizations that can be
found by examining the rather heterogeneous types of invariant information that
cut across traditional conceptual metaphors. These, I argue, can be empirically
investigated, at least in part, by re-examining existing conceptual metaphors,
especially where detailed (ontological) correspondences have been proposed
between source and target domains, and looking for commonalities.
This search for commonalities amongst traditional Conceptual Metaphors can
be described as a meta-analysis of existing domain to domain mappings similar
and complementary to that performed by Grady (1997a,b). Grady also
abandoned the centrality of complex domains in favour of “primary”, concept to
concept mappings. These cut across the traditional domain mappings and
interact, blend or unify with each other to motivate some, but crucially not all,
of the mappings underlying the complex domain to domain mappings. The work
reported here differs from that of Grady in that the mappings sought are
invariant ones.4
In conclusion, I wish to argue that looking for significant generalisations
concerning metaphor interpretation by aligning ontologically complex domains
may be mistaken. Instead, the correct place to look for generalisations is in the
type of information that can transfer in an invariant manner between source and
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
target and conversely in the type of information that is typically ignored. In
section 5, I shall re-examine from the perspective of a putative set of types of
information that partake in invariant transfers, many of the mappings motivating
Jäkel’s (1995) MENTAL ACTIVITY IS MANIPULATION conceptual
metaphor. However, I wish to do more in this paper than merely propose analternative to conceptual metaphor theory or a reorientation towards an
examination of metaphor in terms of the dimensions of similarity between
source and target.5 I would argue that there are a number of problems with the
traditional domain to domain approach which would seem to favour the
approach outlined in this paper. Thus, I shall first look at some of the problems
with using domains to circumscribe the space of possible mappings. I shall then
discuss the importance of what have been described in the ATT-Meta approach
to metaphor understanding (see Wallington et al, 2006; Barnden et al, 2004;
Barnden, 2006; Barnden, 2008) as “Map-Transcending Entities” i.e. source
domain entities, relations, attributes and so on for which it is hard if not
impossible to give a corresponding target domain interpretation, yet which play
a crucial role in allowing the interpretation of the metaphorical utterance to be
inferred.
A final note needs to be made for this introduction, before I discuss certain
problems with domains. By emphasising the role of inference in metaphor
interpretation it should not be assumed that I dismiss the notion of
“conventional metaphors”. Firstly, the same sequence of inferences repeated
often enough should soon become an almost automatic process, triggered uponhearing the metaphor. Secondly, a child hearing what an analyst might classify
as a metaphor may attempt a process of inferencing to derive the meaning, but it
is also possible that she makes a direct sound-meaning pairing. Thus, a final
stage in the process of conventionalisation would be if the sequence of inference
steps were replaced by a single step and the metaphor becomes lexicalised. In
short, a metaphor may become conventionalized either over the history of the
individual user or over the history of the language, with a child lexicalising
what had previously involved a sequence of conventional and automatic
inferences. Furthermore, factors influencing which metaphoric phrases become
conventionalised and which do not presumably include the type of lexical and
collocational conventions that are best explored using corpus examinations of
the type undertaken by Deignan (2005). Nonetheless, I would argue that even
with highly conventional examples, we can see evidence of the role of
inferencing and invariant transfers.
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
transitively to (all) upper domains even though the upper domains do not
directly provide the background necessary to understand the concept? If this is
the case, it makes the definition of a domain particularly inclusive and makes it
correspondingly harder to utilise the concept of a domain to circumscribe the
space of possible domain to domain mappings.
Given this problem, if we wish to keep the concept of a domain in metaphor
theory, it might be best to choose the most restrictive definition and limit a
domain to that against which a concept can be profiled. But does this apply in
practice to proposed domain to domain mappings? To give a concrete example,
one of the phrases Lakoff lists as illustrating the LOVE AS A JOURNEY
conceptual metaphor is: “It’s been a long bumpy road.” According to the
mapping Lakoff provides, “difficulties in the relationship” correspond to
“impediments to travel.” It would seem reasonable to infer impediments from
bumps, so we might ask whether specific bumps, as opposed to the more general
obstacles, are part of the domain of journeys. Now, bumps may be profiled
against the domain of roads (as well as against numerous other surface
domains), and roads may be profiled against journeys. We therefore have an
analogous case to that of the arc, circle and geometric shape. So somewhat
indirectly bumps can be linked to a domain of journeys. But does the existence
of this inferential link entail that bumps is in the domain of journeys and is it
this domain membership that allows bumpy roads to be used to talk about love
affairs via the LOVE AS A JOURNEY conceptual metaphor? Or is what is
important just the fact that bumps can be inferentially linked to the notion of journey, conveying information about the lack of progress? Is the notion of a
domain something of a red herring here?
However, consideration of the bumps examples raises a more fundamental
problem. Much of the information conveyed by a metaphor does not seem to be
specific to any one domain, at least at a degree of abstraction, and conceptual
metaphor theory has always agreed that mappings should be stated at the more
abstract, superordinate, level, rather than a basic level (see Lakoff 1993:211-2).
For example, it can be inferred that bumps/obstacles prevent or disenable the
smooth running of the journey.6 But an enablement/disenablement relationship
is not restricted to the domain of journeys. For example, I would argue that in
the following sentence, which utilises the conceptual metaphor THEORIES AS
BUILDINGS, or perhaps the more general, VIABILITY IS ERECTNESS (see
Grady, 1997a,b), it can be inferred that the purpose of the buttressing is to
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
prevent or disenable the theory-construction from being refuted or toppling, or
conversely to enable it to stay erect.
(2) Later, Freud did locate a couple of key historical documents which
buttressed his argument .
And the notion of a general mapping is not restricted to disenablement. To give
another example, Lakoff (1993) argues that in the following song lyric the
emotion of excitement is inferred and transferred to the target.
(3) We’re riding in the fast lane on the freeway of love.
However, emotional states such as excitement are not only inferred in the
domain of journeys. Finally, we might note that in the bumpy road example,
bumps might be expected to cause physical discomfort. This physical
discomfort might itself lead to emotional discomfort, or the physical discomfort
might stand metonymically for emotional discomfort; consider for example,
“heartache”. In either case, emotional discomfort in the source corresponds to
emotional discomfort in the target. Numerous other examples will be discussed
of domain-neutral information being inferred from metaphorical utterances.
The arguments so far against the utility of domain in metaphor theory for
anything other than superficial cataloguing have been, firstly, that having
domains does not seem to constrain possible inference chains in any natural wayand, secondly, that much of importance in the interpretation of a metaphor
seems to apply to numerous metaphors and is not profiled against any specific
domain. I now wish to consider concepts which receive a standard non-
metaphorical interpretation by being profiled against a domain and so should
clearly be thought part of the domain, yet which seem to lead to uninterpretable
metaphors when used in what appears otherwise to be a conventional and
productive source domain to target domain mapping. Thus both Grady
(1997a,b) and Clausner and Croft (1997) have discussed the conceptual
metaphor THEORIES AS BUILDINGS and noted the non-occurrence of such
metaphors as the following:
(4) That theory lacks French windows.
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
The existence of the right of the mind gap would lead Grady to conclude that
MIND AS PHYSICAL SPACE is not a primary metaphor and it may be that
further refinement can eventually come up with a set of primary metaphors for
which there exist no gaps. However an issue that bears on the nature of gaps,and which is often over-looked, is that even for entities which can be used
felicitously in a metaphorical utterance not every inference that one might make
about the entity using our normal common sense knowledge of the putative
source domain is acceptable. For example, Mussolf (2004) gives examples in
which the European Union is described as a marriage. We might assume that
such metaphors are licensed by a UNION AS MARRIAGE conceptual
metaphor. But this would be a marriage with -currently- 27 partners, which
would certainly be illegitimate in all the European Union countries. But the
illegitimacy of the European Union is presumably not an inference that the user
would expect to be drawn. Consider the utterance this policy reflects the views
of the majority. Now, a reflection inverts the image and so the utterance perhaps
ought to be taken counter-factually to mean that the policy is opposed by the
majority. But of course, it doesn’t. It is not as if it violates our common sense
knowledge of mirrors; we know that mirrors invert and even utilise the fact in
such metaphorical utterances as this is the mirror-image of what was wanted . In
short there appear to be inference gaps as well as the more normal entity gaps
What this suggests is that it might not be certain entities, such as windows or
right sides, per se that lead to illegitimate metaphors. Instead, it is the drawingof certain inferences from these entities that connect to the core correspondence.
It is these that are impossible or perverse given the context. With acceptable
entities, on the other hand, inferences, which are likely to be made by the
producer of the metaphor and followed by its interpreter, are possible. Thus, to
return to an earlier example in which the felicity of the novel phrase: You would
need an electron microscope to see the point in that was noted, we can infer,
given our knowledge of what electron microscopes are used for, and assuming a
conceptual metaphor COGNIZING IS SEEING, that the degree to which
something can be seen and hence understood is very low. A similar ease of
inferencing occurs in the case of back of the mind and, the less conventional,
edge of the mind . I will argue, following the broad lines of the ATT-Meta
approach to metaphor interpretation (see, Barnden, 2006; Barnden, 2008;
Barnden et al., 2004; Wallington et al., 2006), that for someone not at the
periphery, it can easily be inferred that peripheral areas can only be interacted
with to a very limited degree, and it is this limited interaction with, or awareness
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
of, the idea that represents the intended message of these sentences..
Conversely, it is unclear what inferences one could draw within the context of
THEORIES AS BUILDINGS about the relation of French windows to the rest
of the building. A similar point can be made about why an interpretation of
utterances such as it lay on the right of her mind in the context of MIND ASPHYSICAL SPACE is absent. Its absence makes sense if one tries to imagine
what inferences could be drawn from its use, and, in particular, whether the
conclusions of such inferences represent the type of information that typically
transfers in metaphor.
To conclude, although it may be possible to give precise definitions of domains,
there are reasons to doubt that the notion of domain used in discussions of
cross-domain correspondences can be given a precise enough definition to
ground the notion of source-target transfers. Furthermore, much of what is
transferred is of an abstract nature that would transcend any domain boundary.
Consequently, I take a much more open-ended approach, with potentially
anything that can be inferentially linked to a central source-target mapping
being possible to incorporate into a metaphorical utterance.10 But on the face of
it this would fail to account for the apparent systematicity unless there is some
means of controlling the possible inferences. The control cannot be exerted by
domain boundaries. The proposal I wish to advance in this paper is that the
notion of a correspondence or mapping between two ontologically complex
domains should be replaced by that of a core mapping and the assumption that
only information of a certain type typically transfers. Such information will beinferred as holding of the situation described by the metaphorical utterance if
the producer and interpreter of the utterance in its context can follow the
necessary chain of inferences to the relevant abstract information types.
3 Correspondence Approaches and Map-Transcending Entities.
The previous section argued that it is a mistake to account for systematicity by
appealing to mappings or correspondences between well-defined domains. In
this section I shall aim towards the same conclusion. However I shall do this not
by querying the nature of domains, but by pursuing one of the principle
theoretical claims of ATT-Meta theory (Barnden, 2006; Barnden, 2008; Barnden
et al., 2004, Wallington et al., 2006) and arguing that many of the entities,
relations, attributes and so on used metaphorically in an utterance to talk about a
particular target need have no actual target correspondents.
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
What can be made of this pattern of inferences that allows us to conclude that
the vehicle is not moving towards its destination? Both the vehicle and the
destination have correspondents in the target domain of the love affair, namely
the love relationship itself and the lovers’ common goals. With thesecorrespondences, it might be assumed that the source domain conclusion can be
transferred to become a target domain conclusion. However, this is the transfer
of the conclusion. The conclusion would never have been reached without the
premise that something -the spinning wheels- was causing the vehicle not to
proceed. So what is the target domain correspondent of the spinning wheels,
whose presence is needed in order to allow the entire inference pattern, as
opposed to just the conclusion, to transfer between source and target? Nothing
in the list of four ontological correspondences would seem to be appropriate.
And note, furthermore, that our conclusion involves a negation: not moving
towards a target. The negation transfers to the target, but there is no formal
discussion of why this should be so and whether negation in the source always
corresponds to negation in the target.
A look at the other examples listed finds similar cases where the lack of a target
domain correspondent would prevent crucial aspects of the source domain
inference pattern from mapping to an inference pattern about the target. For
example, what is the target domain correspondent of the rocks in the utterance
our marriage is on the rocks? A similar point can be made about the location off
the tracks in our relationship is off the track . But, as before, these statementscan be easily interpreted. Both of them would seem to permit the same
conclusion that was reached about the spinning wheels, namely that the
relationship/vehicle is not progressing towards the destination and hence not
towards the lovers’ common goal, and that consequently the relationship is not
functioning as it ought. But without target domain correspondents for these
items the conclusion will not transfer if it is inference patterns that map between
source and target.
Note that it doesn’t seem quite right to assume that the spinning wheels, rocks
or lack of tracks are impediments in the sense of the third of Lakoff’s
correspondences. Whilst one might assume that the bumps in the statement it’s
been a long bumpy road are impediments and point to specific difficulties that
have occurred over the course of the relationship, there does not seem to be an
assumption with the other cases we have discussed that there are specific causes
to the relationship not progressing and hence functioning. Whatever wheels
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
might refer to, it is a vague and very unspecific target domain entity.
Furthermore, there is also the point made in the previous section that a
relationship described as a bumpy road suggests that there was some emotional
discomfort, an inference that would not appear to follow from the assumption
that there was an impediment.
In short, a careful study of these examples supports the view of ATT-Meta
theory that there can be entities referred to in a metaphorical utterance that need
have no correspondent or mapping in any target or literal interpretation of the
utterance. Instead, their role is in allowing certain conclusions concerning the
source to be drawn to be drawn and it is these conclusions that transfer, not
necessarily all the items that were involved in the drawing of the conclusions. 11
Another sentence often used to illustrate the LOVE AS A JOURNEY conceptual
metaphor is: We’re at a crossroads. Stating that we are at some location on a
road might be taken to imply that we are going somewhere along the road and
hence have a destination. This would correspond to a target domain common
goal . However, there is no specific correspondent given for crossroads. The
source domain inference that if one is at a crossroads, then there is a choice of
possible destinations cannot transfer to the target becoming the inference that
there alternative relationship goals, if, as is stated, inference patterns are
mapped from one domain to another.
Now a possible solution to the problem might be to assume a lacuna in thecorrespondences listed above and add a fifth correspondent that could be used to
interpret crossroads; something along the lines of alternative destinations
correspond to alternative life-goals. However, the mapping already lists as
correspondence 3, life goals corresponding to destinations. It might be thought
more parsimonious to factor out from this putative correspondence the
commonality that could be dealt with by correspondence 3. This leaves
alternatives mapping to alternatives. But, this is just the type of invariant
transfer I suggested in the introduction underlies much metaphor. If further
evidence shows that alternatives inferred in the source always map across to
become alternatives in the target regardless of which conceptual metaphor is in
play, then it would be redundant and missing a generalization to include a
specific correspondence as part of the LOVE AS A JOURNEY Conceptual
Metaphor.
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
I would argue that a similar factoring out can be performed with the fourth
correspondence: difficulties in the relationship correspond to impediments to
travel , if it is first rephrased as: Factors that prevent the relationship from
progressing are factors that prevent the vehicle from travelling . Now, since we
have a relationship (progressing towards a goal) corresponding to a vehicle(travelling towards a destination) listed as correspondence 2 (and 3), what is
particular to the rewritten correspondence 4 is factors that prevent
corresponding to factors that prevent .
I will follow ATT-Meta theory and assume that ability, inability, enablement,
disenablement, letting and prevention (more broadly, causation) tend to map
invariantly. I will assume that an inference can be made that the journey is being
prevented from proceeding as it should. This, a reader will understand in reality
as concerning a love affair rather than a journey, because of the core mappings
of the LOVE AS A JOURNEY conceptual metaphor and she will further
understand in reality that the affair isn’t progressing, because of the
disenablement adjunct to the LOVE AS A JOURNEY mapping that allows
disenablement in the source to be viewed as disenablement in the target.
In conclusion, and in agreement with the view of the ATT-Meta approach, there
often appears to be a lack of parallelism or correspondence between source and
target. Furthermore, where correspondences are stated as in Lakoff’s LOVE AS
A JOURNEY correspondences, they are often little more than rewordings of the
source terminology in order to transfer an invariant property, as with difficultiesin the relationship corresponding to impediments to travel . All this creates
problems for the assumption of domain to domain mappings, but if, instead,
what transfers are conclusions of a certain type and if the purpose of the source
entities in the utterance for which there are no correspondents is to allow
connections to be inferred with central and invariant correspondences and
append what information they enabled to be inferred to these correspondences,
then the lack of parallelism would be expected.12
4 Invariant Transfers
The two preceding sections have suggested that it is a mistake to look for the
systematicity of metaphor in correspondences between conceptual domains. In
section 2, I argued that the notion of domains is problematic, at least with
respect to circumscribing possible metaphor mappings, being both difficult to
delimit and needing to contain elements such as excitement, degree and so on,
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
which presumably could be inferred from numerous other unrelated
metaphorical utterances. In section 3, I criticized the claim that inference
patterns must transfer from source to target by casting doubt on the notion that
entities, relations, attributes and so on referred to in a metaphorical utterance
must have target correspondents.
So what universal types of invariant information can be inferred? It does not
seem to be the case that information of any type can be inferred, even if there is
strong contextual support for the drawing of such an inference. For example,
Hobbs (1990:59,60) noted the word elephant used metaphorically can convey
different information depending upon the context.
(8) Mary is graceful, but John is an elephant.
Patricia is small, but James is an elephant.
Susan forgets everything, but Paul is an elephant.
Jenifer is subtle, but Roger is an elephant.
However, the colour red is surely as salient an aspect of tomato or pillar-box as
clumsiness is of an elephant, yet the following metaphorical utterance doesn’t
seem to work.
(9) Mary’s pencil is blue, but John’s pencil is a tomato/pillar-box.
And consider the well known Juliet is the sun. It might be concluded that colour is not the type of information that typically accompanies a core source-target
mapping.13
I have already made some suggestions as to what types of information may
transfer invariantly, such as information concerning enablement and
disenablement, the degree to which something holds, and to emotional states.
However, I have not yet shown that the examples in which such information is
inferred are representative of a more general phenomenon. It might be the case
that the types of inference that transfer invariantly differ according to the
particular conceptual metaphor chosen. Indeed, the class of GENERIC IS
SPECIFIC metaphors (see Lakoff and Turner 1989) or Grady (1999)), such as:
LONG TERM PURPOSEFUL ACTIVITIES ARE JOURNEYS, where a
journey is a specific instance of a long term purposeful activity, constitute such
a case.14
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
However, the claim I wish to support in this paper is that there exists a set of
invariant mappings which have more universal application –at least by default.
In the ATT-Meta approach this set of invariant mappings have been called View-
Neutral Mapping Adjuncts (VNMAs); adjuncts because they accompany, or are
an adjunct to, traditional source-domain to target-domain mappings, and view-neutral because they can apply regardless of which metaphorical view one is
taking.15 In what follows, I shall list of a number of different types of
information which seem to get transferred between source and target by these
VNMAs. In each case, I shall also attempt briefly to motivate the particular
VNMA by giving a few examples of the invariant transfer of the type of
information being discussed using a range of what might be thought of as
different source and target domains. The list is not intended to be complete (see
Wallington and Barnden, 2006 for a fuller list with more discussion. See also
Wallington et al., 2006 and Barnden et al., 2003). In addition, since any
metaphorical utterance typically conveys a mixture of information, the
examples given do not pretend to consist of all, or even the most prominent
type, of the information being conveyed.
CAUSATION/ABILITY/FUNCTION VNMA.
Causation, ability, function, prevention, helping and (dis) enablement
relationships between events or other entities in the source map to causation,
prevention, etc relationships between their mappees (if they have any).
A couple of examples that have utilised the invariant transfer of enablement (or disenablement) have already been discussed. Thus, in sentence 2, repeated
below, I argued that the buttressing documents enabled Freud’s arguments to
stay erect. This enabling relation between the documents and the arguments
transfers to the target as an adjunct to the core mapping between THEORIES
and BUILDINGS.
2 Later, Freud did locate a couple of key historical documents which
buttressed his argument .
Similarly, I argued that the relevance of bumps in Lakoff’s LOVE AS A
JOURNEY conceptual metaphor was that they prevented the love affair-journey
from progressing as it should.
QUALITATIVE DEGREE VNMA.
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
This involves correspondences between ISSUES and PHYSICAL OBJECTS
and between MENTAL ACTIVITY and PHYSICAL INTERACTION.21 The
fact that the object can be manipulated/viewed in more than one way, as can beinferred by the claim that there are different sides, will be transferred to the
target accompanying the MENTAL ACTIVITY IS PHYSICAL INTERACTION
mapping.
EVENT-SHAPE VNMA.
Aspectual features of events/situations/processes, such as whether they have a
start or end, or are intermittent, map identically to mappee events/ situations/
processes.
All episodes, including source and target, need an event-shape. By default the
event-shape of the target will be inherited from the source.22 Compare for
example:
(20) He was searching for an answer to the question.
(21) He wanted an answer to the question.
The first example denotes an activity, the second a state, and such information
concerning the event will be mapped by the EVENT-SHAPE VNMA. No
specific target correspondent is required for the searching or wanting .
MENTAL/EMOTIONAL STATES VNMA.
If some agents in the source domain map to some agents in the target domain,
then mental and emotional states of the agents map identically, except that their
objects or propositional contents (if any) are modified suitably by any mapping
relationships that apply, and provided that this modification can be done. This
VNMA also transfers any emotional state induced in the users of the metaphor
by the source domain description to the target.23
(22) Your room is a cess-pit.
This utterance plausibly relies upon the mapping STATES ARE LOCATIONS
(state of your room is a location-cess-pit). The disgust evoked by the use of the
term cess-pit is conveyed by the MENTAL/EMOTIONAL STATES VNMA.
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
46. The intrigues of the court had sharpened her wits.
47. He used extensive reading to hone his intellect.
However, let us use instead the familiar metaphor ANALYSING IS CUTTING.
I can then assume a source domain inference that the degree to which one cancut something, and hence analyse it, is high if the tool is sharp. This will be
conveyed by the DEGREE VNMA
A somewhat similar analysis can be given to utterances such as:
48. This is a hard nut to crack
But in this case the idea-object or nut is being profiled rather than the cutting or
cracking implement.
The following utterances about the MIND-TOOL can be analysed as allowing
the inference that the MIND-TOOL is not functioning properly:
49. He flies off the handle quite easily.
50. Her mind was not on it
This lack of function will be conveyed by the CAUSATION-ABILITY-
FUNCTION VNMA.
Jäkel uses the following utterance to motivate his MIC 4d, SOLVING A
PROBLEM IS TAKING THE SOLUTION-CONTENTS OUT OF THE
PROBLEM-CONTAINER.
51. We dredged up a load of sordid facts about her.
The notion of facts-object being brought to a position where they can be easily
operated upon or interacted with is similar to many other examples I have
analysed. However, note the negative value-judgment conveyed by the word
dredge. I shall assume this is conveyed to the target by the VALUE-
JUDGMENT VNMA.
MIC 5 is FORMING IDEAS IS SHAPING RAW MATERIALS. This differs
from other examples in that the idea-object is produced or manufactured rather
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
candidature is subject to the constraint that the speaker must assume that the
hearer will draw the correct inferences and, most importantly, that the
information so inferred belongs to one of the types of information that can
transfer invariantly between source and target as an adjunct of the core
mappings. Following the ATT-Meta approach, I put forward a set of VNMAsthat characterise the types of information that appear to map in an invariant
manner. It must be stressed that determining the set of VNMAs is an empirical
exercise, which requires careful examination of a wide range of metaphorical
utterances in order to factor out common invariant properties.
The claim made in this paper, however, consists of more than that there is an
alternative means of explaining the systematicity found in metaphor. I argued
that there is no operational means of defining source domains that could
usefully be used to delimit the set of possible metaphors and I noted that much
of what is transferred would not normally be characterised as belonging to just
one particular source domain. I also noted that many of the entities, relations,
attributes, and so on in particular metaphorical utterances do not have any
correspondents in the target. Thus I would claim that there are empirical
problems with domain to domain mapping approaches. In general, it is not
entities that are important, or even how different entities relate or are structured,
but rather the types of conclusion that can be drawn from their use.
It could be argued that the proposal is more of a shift in emphasis in doing
metaphor research. Others have noted the importance of invariance in metaphor.Indeed a traditional claim is that metaphor involves searching for a similarity
between the source and the target, although ATT-Meta’s assumption that
correspondents need not be required for many source entities argues against too
superficial a similarity. This lack of source-target parallelism also argues against
the claim that metaphor is a type of analogy or is strongly related to analogy
(see Gentner et al., 1988). If we turn to specific approaches to metaphor and
analogy, further differences from the approach presented here can be found.
Thus, approaches based on Structure Mapping Theory (Gentner 1983) argue that
systematic relations and the entities involved in them will transfer between
source and target, whilst one-place relations such as attributes will not. Such an
approach may well account for the CAUSATION-ABILITY-FUNCTION
VNMA. However, I also assume that VALUE-JUDGMENT attributes that
pertain to something will also transfer by default, as will DEGREES.
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
1 I shall follow the convention of capitalising source-target mappings and continue the process both when the mappings are
from concept to concept, rather than domain to domain and also for types of information that are mapped in an unchanging
or invariant manner between source and target.2 But unlike Relevance Theory where the interpretation of a metaphorical utterance is broadened or narrowed until its
relevance to the discourse is able to ground the inference process, I will argue that what grounds the inference process is
whether information of a specific type that is relevant to the discourse can be inferred.
3 This analogy implies no commitment to Chomskyan linguistics. The importance of constructions has recently received a
great deal of support from linguists working in a number of different theoretical approaches(e.g. Goldberg 2006, Sag et al.
2003).4 I also make no assumptions concerning the embodiedness of the concepts involved.
5 Using the term conceptual metaphor theory overlooks the fact that there have been major changes and advances in thetheory over the years. See Lakoff (2008).
. See Barnden 2008 for a discussion of some of the problems such a view of domain would pose for metaphor theory.
6 According to the online edition of the OED, the meaning of “disenable” is: To render unable or incapable; to disable: the
reverse of enable.
7 A phrase motivated by ORGANIZATION IS PHYSICAL STRUCTURE that is not about theories might be: “the fabric of
society”. A phrase motivated by VIABILITY IS ERECTNESS that is not about theories might be: “This situation will not
stand”.
8 For reasons such as these, Grady abandons the notion of a domain for what he calls “primary metaphors” and assumes
instead concept to concept mappings, relying on a somewhat sketchy notion of the unification and/or blending of primary
metaphors to account for metaphorical systems such as THEORIES ARE BUILDINGS.
9 Examples were found in which “off” had been misspelled as “of” and the words occur within the larger phrase “I have notlost the birth-right of my mind” in Act 3, scene 1 of John Dryden’s play Aureng Zebe.
10 Note that to define a domain as any entities, relations, attributes and so on that can be inferentially linked to a core
element leads to a circularity.
11 In ATT-Meta theory, entities that are not directly part of a mapping, either a pre-existing mapping such as the
correspondences in the LOVE AS A JOURNEY mapping, or a mapping that needs to be created on the fly, are termed
“Map-Transcending Entities” or MTEs. One of the principle theoretical claims made by ATT-Meta is that very little source
to target mapping needs to be done on the fly.12 We might go further than stating that unmapped source entities allow inferences to be made to stating that they initiate
the inference process. In the ATT-Meta system this is enshrined in the principle of “Transfer-Warranting by Unmapped
Structure” (see Wallington et al 2006, Barnden and Lee 2001).
13 The accompaniment ought to be emphasised here. I am not claiming that a colour cannot be the target of image
metaphors. An added complexity is that the presence or absence of a determiner seems to make a difference. “John’s is
pillar-box” seems marginally more acceptable to me.14 The importance and widespread use of GENERIC AS SPECIFIC metaphors is often overlooked, the GENERIC AS
SPECIFIC being disguised by the terminology used to describe the source and the target. Thus, Grady decomposedTHEORIES AS BUILDINGS into PERSISTING IS REMAINING ERECT and ORGANIZATION IS PHYSICAL
STRUCTURE. But “persisting” is a close synonym of “remaining” and a “structure” is a type of “organization.”
Consequently, the two metaphors could be reworded as: PERSISTING IS PERSISTING IN AN ERECT STATE and
ORGANIZATION IS PHYSICAL ORGANIZATION. If this is done, the GENERIC IS SPECIFIC structure is brought out.
15 The term COMMAs (Conceptual Metaphor Mapping Adjuncts) was used instead of VNMAs in Wallington et al., 2006.
16 This sentence and the ensuing discussion is related to an ATT-Meta example, see for example, Barnden 2008.
17 This is essentially Jäkel’s (1995) MIND AS A WORKSHOP metaphor.
18 There s an alternative, but very similar analysis of these sentences and that involves replacing the MENTAL ACTIVITY
AS PHYSICAL INTERACTION (WITH AN OBJECT) mapping with COGNISING AS SEEING (AN OBJECT), and
inferring from the phrase “the far reaches” that the idea-object can barely be seen. In fact, we might collapse the twodifferent mappings into the more general MENTAL ACTIVITY AS INTERACTION WITH AN OBJECT THROUGH THE
SENSES, with the former involving the sense f touch and the latter, the sense of vision. This, of course suggests that the
senses of hearing, smell, and taste can also be used to describe mental activity. I will not pursue the idea further here.
19 Note the existence of the DEGREE VNMA at work as well.
20 This VNMA frequently operates in conjunction with the Mental and Emotional States VNMA, to be discussed shortly,
allowing the inference of a choice.
21 Or possibly COGNIZING AS SEEING; it is often difficult to determine whether the tactile or visual senses are being
utilised.22 Glaseby (2003, 2007) argues that with idioms that take the form of fake resultatives, such as “She painted the town red,”
the event shape of the literal interpretation is telic: it is an accomplishment and has an end point, with the town painted red.
However, the event shape of the idiomatic interpretation is that of an activity. Thus, one can say “they painted the town red
7/27/2019 Systematicity in Metaphor and the Use of Invariant Mappings - Wallington-ApresRAAM06-19feb09
for a few hours” only under the idiom interpretation. In short the event shape does not appear to transfer invariantly.
23 I assume that mental and emotional states are frequently combined and interconnected, making it difficult to talk of pure
mental or emotional states. It may even be the case that there is a cline between mental and emotional states of mind.24 As with the DEGREE VNMA, RATE must be interpreted qualitatively rather than in terms of some absolute rate, for
example as: slow, fast or at a medium rate, and inferred using everyday, common sense, knowledge about the
metaphorically cast object. For example, the following sentence was found via an internet search: “Evolution occurred at a
glacial pace” (www.straight.com/print/118354). It can be inferred that glaciers move extremely slowly and this very low
degree is what is transferred to the rate of change of evolution. But of course, glaciers move at many times the order of
magnitude quicker than evolution proceeds.25 This is broadly the same as the core metaphor I have called MENTAL ACTIVITY IS PHYSICAL INTERACTION.
26 It should be noted that space permits consideration of only a selection of his examples and I shall usually discuss onlysome of the inferences that may be drawn, but see Wallington and Barnden (2006) for a fuller analysis.
27 In fact Grady (1997a) motivates the IMPORTANCE IS WEIGHT primary metaphor by just this type of inference.
28 See Barnden’s Metaphors of Mind database http://www.cs.bham.ac.uk/~jab/ATT-Meta/Databank/) for discussion of the
relation between MIND AS A CONTAINER and MIND AS PHYSICAL SPACE.
29 An anonymous reviewer has questioned “razor wit”. The following is from an internet search: Sexual perversity in
Donegal. With whiskey, suicide and razor wit , the Dublin Theatre Festival is in fine form.