From three dimensions to five steps: The value of deliberate metaphor1
Gerard Steen, Amsterdam (g.j.steen@vu.nl)
Abstract
Recent interest in the nature of deliberate metaphor has demonstrated the need for a three- dimensional model of metaphor. Metaphor is not just a matter of language and thought, but also of communication. How can this starting point be connected with a discourse-psychological framework for metaphor in which it is assumed that people have to represent metaphor as part of a number of complementary mental models for discourse? And how can this discourse- analytical approach be formalized in the five-step method for metaphor analysis going from language to thought: Can this method be adjusted to cater to the formal analysis of metaphor in communication as well, so that a distinction can be made between deliberate and non-deliberate metaphor in the five-step method? These are the questions addressed in this article.
Das jüngste Interesse an der bewusst eingesetzten, gezielten Metapher hat die Notwendigkeit eines dreidimensionalen Metaphernmodells verdeutlicht. Metaphern sind nicht nur eine Ange- legenheit von Sprache und Kognition, sondern auch für die Kommunikation. Dieser Beitrag möchte folgende, sich aus diesen Voraussetzungen ergebene Fragen beantworten: Wie lässt sich dieser Standpunkt verbinden mit einem diskurs-psychologischen Ansatz, in dem vorausgesetzt wird, dass man sich die Metapher als einen Teilbereich komplementärer, im Diskurs aktivierter mentaler Modelle vorstellen soll? Und wie kann dieser diskursanalytische Ansatz in einem fünf- schrittigen Modell, das die sprachliche und kognitive Ebene umschließt, formalisiert werden? Kann dieses Modell auch die kommunikativen Dimensionen erfassen, so dass klar zwischen ge- zielter und ‚ungezielter’ Metaphernverwendung unterschieden werden kann?
1. Introduction
Some metaphors are used deliberately as metaphors, whereas most metaphors are not. When Shakespeare asks, ‘Shall I compare thee to a summer’s day?’, he is entertaining the thought whether or not he will set up a metaphorical comparison between his beloved addressee and a summer’s day. In other words, he is attend- ing to metaphor as metaphor and is about to make a decision about his intentions for his next move: ‘Thou art more lovely and more temperate’.
It is true, of course, that we are not talking about the real Shakespeare but about his persona in the sonnet cycle. Yet this is a well-known example that can easily
1 I am very grateful to useful comments by Tina Krennmayr and Gudrun Reijnierse.
83
metaphorik.de 21/2011
stand for what can happen in regular language use. Examples of deliberate meta- phor from real life may be found in the language of politicians, instructors, sports commentators, and so on (Charteris-Black, 2004; Cameron, 2003; Beger, 2011).
Deliberate metaphors differ from non-deliberate metaphors in that they involve mandatory attention to the fact that they are metaphorical. This is not meant to imply that addressees have to be able to label a metaphor as a metaphor for it to count as deliberate, though. That is a metalinguistic skill that is irrelevant. In my view, a metaphor is deliberate when addressees must pay attention to the source domain as an independent conceptual domain (or space or category) that they are instructed to use to think about the target of the metaphor.
Deliberate metaphors are constructed in such a way that the sender invites or sometimes even instructs the addressee to in fact set up a cross-domain mapping in their mental representation of the discourse in order to view one thing in terms of something else. Consider famous Dutch football coach Rinus Michels who said
´Football is war´. His utterance was intended as an invitation for his audience to pay attention to football from the independent standpoint of war, and to draw relevant conclusions (implicatures) from that figurative comparison. Deliberate metaphors are perspective changers.
This is not what happens with non-deliberate metaphor. For instance when peo- ple talk about arguments in terms of war or life in terms of a journey, they do not necessarily have to attend to the difference as well as relation between the two respective domains that may be distinguished in semantic and conceptual struc- tures by cognitive linguists and other metaphor analysts. An utterance like
´Lakoff attacked Glucksberg´ is typically meant to be understood as an utterance
involving just one conceptual domain, the target domain of academic debate. Neither the sender nor the addressee is supposed to pay any attention to any is- sues that are related to physical fighting and war as they are representing this ut- terance as part of the meaning of the discourse. Even if they continue to talk about argument in terms of war, this does not necessarily mean that they activate their knowledge about war in order to think and talk about debates. They may just employ polysemous words that are conventionally used to talk about de-
bates; these may in the past have been metaphorically derived from the domain
84
Steen, The value of deliberate metaphor
of war, but it is quite possible that at present they may have lost that connection in people’s individual minds.
The possibility that a lot of metaphor may be processed non-metaphorically con- stitutes a paradox (Steen 2008). To some it has raised the question whether ‘La- koff attacked Glucksberg’ can even count as metaphorical. I think it does, for it is still possible that the source domain of fighting or war is activated in the proc- esses preceding the representation of the sentence in working memory, as has been vociferously contended by cognitive linguists and as may be supported by a range of experimental studies (Gibbs 2011a and c). If we abide by the conceptual definition of metaphor as a cross-domain mapping, linguistic expressions such as
‘Lakoff attacked Glucksberg’ have to be kept on board as potentially metaphori- cal, since they may evoke some form of mapping across conceptual domains. However, the processes leading up to the product of metaphor comprehension in working memory (or attention) are a separate issue that are largely immaterial to the question of whether their product counts as a deliberate metaphor or not—I will argue here the latter issue turns on attending to the source domain as part of the represented meaning of the utterance.
The distinction between deliberate and non-deliberate metaphor is controversial. In a recent exchange, an attempt was made by Gibbs (2011a), Steen (2011a), Deig- nan (2011), Müller (2011) and Gibbs (2011b) to clarify what it means to call a me- taphor deliberate or non-deliberate. As a result of that exchange I now hold that deliberate metaphor is not necessarily the same as conscious metaphor (Steen
2011a): I propose that deliberate metaphor gets represented as metaphor in atten- tion, and that it thereby affords conscious metaphorical cognition, but that con- scious metaphorical cognition does not necessarily have to occur (Steen in press). Also, deliberate metaphor should not be conflated with novel metaphor, non- deliberate metaphor being conflated with conventional metaphor: conventional metaphor can also be used deliberately, as happens when newspaper headlines use conventional metaphors deliberately for the purpose of punning and humor and drawing the attention of the reader. At the same time, though, there still re- main many questions about deliberate metaphor, for instance the possible asym- metry between production and reception: a metaphor may be deliberately pro-
duced as a metaphor but not received as one, or, the other way around, a meta-
85
metaphorik.de 21/2011
phor may have been produced non-deliberately as a metaphor but still be re- ceived as a cross-domain comparison by the interlocutor or audience. Careful theoretical and empirical work will have to elaborate how the notion of deliberate metaphor can be conceived of in these contexts.
Deliberate metaphor therefore has two kinds of values: firstly, it is a special kind of metaphor use which has its particular values to language users, in that it draws their attention to the perspective of the source domain; and secondly, it is a special kind of phenomenon which has its particular values to practitioners of the academic study of metaphor, in that it reveals that metaphor is not just a matter of language and thought but also of communication. In this paper I would like to try and demonstrate both kinds of values. I will begin with the empirical and theoretical issues that have to do with deliberate metaphor as a specific part of discourse: I argue that deliberate versus non-deliberate metaphor is an issue of communication, which is an independent dimension of metaphor in discourse next to the ones of metaphor in thought and metaphor in language (Steen 2008 and 2011b). Then I will continue with a theoretical exploration of the way in which metaphor can be analyzed by means of a five-step method for analysis I proposed some time ago (Steen 1999a; Steen 2009; Dorst 2011; Krennmayr 2011): I will reinterpret and adjust the five-step method for the purpose of catering to the distinction between non-deliberate versus deliberate metaphor, and tie this back to the three dimensions of metaphor that can be distinguished in discourse. My main claim is that deliberate metaphor has been unduly ignored and that it raises fundamental and new questions about structures, functions and processes of fig- uration in discourse, but also about the ways these have been studied.
2. Metaphor and discourse: three dimensions
My proposal for a distinction between deliberate and non-deliberate metaphor (Steen 2008; Steen 2011a; Steen 2011 b; Steen in press) is not original: various scholars before me have used the distinction in roughly comparable ways (Shen and Balaban 1999; Cameron 2003; Charteris-Black and Musolff 2003; Semino
2008). What is original about my proposal, however, and what is crucial for its
further theoretical development and empirical investigation, is my aim to connect
86
Steen, The value of deliberate metaphor
metaphor processing, including deliberate versus non-deliberate metaphor proc- essing, to the psychology of discourse, as represented by for instance Van Dijk and Kintsch (1983), Kintsch (1998), Van Dijk (2008), Graesser, Millis and Zwaan (1997), Perfetti (1999), Macnamara and Magliano (2009) and Graesser and Millis (2011). When metaphor is used, whether deliberately or non-deliberately, it is not used as part of isolated utterances against a background of conceptual systems that inform the meaning of the words in a sentence; instead, metaphors are al- ways part of utterances that are typically small psychological and social acts in encompassing moves and transactions between writers and readers, or speakers and listeners. This requires a three-dimensional model for metaphor, in which its linguistic, conceptual, and communicative aspects are all taken into account. Whichever models for metaphor processing are hence proposed by cognitive lin- guists, psycholinguists and cognitive psychologists, they have to be compatible with such more encompassing and general models of discourse processing as de- veloped in the community of discourse psychologists.
The main issue that needs to be addressed in this context is the need for a multi- level representation of all metaphor in discourse. Discourse psychologists have made a distinction between a number of different but complementary mental models that people construct during discourse processing, of which I have used the following four for developing my approach to metaphor analysis, eventually leading up to the concrete details of my proposal for deliberate metaphor:
1. The surface text, which involves the representation of the linguistic nature of the message exchanged in communication;
2. The text base, which involves a representation of the conceptual nature of the message in the form of a linearly and hierarchically ordered series of idea units, or propositions;
3. The situation model, which is another representation of the conceptual na- ture of the message, this time in the form of a non-propositional format which includes all kinds of referential information;
4. The context model, which is a representation of the discourse situation and the participants involved in exchanging the message, the medium of the message, the message as tool for interaction, and so on.
87
metaphorik.de 21/2011
The surface text captures the linguistic dimension, the text base and situation model, the conceptual dimension, and the context model the communicative di- mension of metaphor. Empirical research has shown that language users need to construct at least parts of these distinct types of mental representations during language use processes to be successful in their communicative behavior (Mac- namara and Magliano 2009; Graesser and Millis 2011).
Metaphor can be connected to this multi-level approach to the psychology of dis- course processing in two ways. I have collected behavioral data which show that aspects of metaphor processing can be usefully interpreted with reference to the distinct mental models for discourse processing (Steen 1994; Steen 2004 and Steen
2006). For instance, I asked participants to rate metaphors from Bob Dylan’s song
‘Hurricane’ on a number of scales for affective value, and found that the same metaphors received different scores on these scales when presented as linguistic expressions versus depictions of a situation versus messages between interlocu- tors (Steen 2004). I believe that these three types of presentation are the three ba- sic dimensions of all metaphor use (Steen 1994; Steen 1999b and Steen 2008). In this particular empirical study, they were meant to be associated with the psy- chological representation of metaphor in a surface text, a situation model, and a context model, respectively. It hence seems possible to experimentally distinguish between the various complementary roles metaphors can play during discourse processing.
The second way in which metaphor can be connected to the multi-level approach to discourse processing is formal instead of behavioral. I developed a formal framework for metaphor analysis which includes five separate analytical steps for the distinct representation formats of metaphor in the multi-level representa- tion of metaphor in discourse (Steen 1999a; Steen 2009). This approach can show how one metaphorical utterance can be analyzed as a potential part of the surface text, the text base, the situation model, and the context model. The latter aspect, metaphor as part of the context model, is a recently developed interpretation of the five-step method, which I will illustrate below, and is closely related to the
value of deliberate metaphor.
88
Steen, The value of deliberate metaphor
Before we can demonstrate the five-step method, however, let us first examine the more general connection between metaphor and discourse, involving the three dimensions of language, thought and communication, and do so by looking at deliberate metaphor. Consider the following data from Beger (2011). It is an ex- cerpt from a university lecture on the nature of aggression as treated by a hy- draulic model developed by Konrad Lorenz. Potentially deliberate metaphors have been identified by Beger and presented in italic and bold print:
“But think about this tank of water as the reservoir within your soul, that aggressive impulses are dripping into. Little hassles and frustrations of day-today life keep adding new bits of aggressive impulses to who you are. And as this tank fills up, the pressure of the weight of these impulses be- comes stronger and stronger and they push on this plug that keeps it bot- tled up. Now, you don’t behave aggressively, until all this stuff kind of ex- plodes and comes shooting out of you. And the weight – and there’s two different factors here. One of the factors in this model as to whether you become aggressive or not, is the weight of the water in the tank – or the amount of aggression that’s piling up. Because as that gets deeper and dee- per, and heavier and heavier, there is more pressure for it to come out. But at the other end, there are stimuli out there in the world that you might be exposed to – aggression-eliciting stimuli, aggressive models, and what have you, that will be pulling on the plug from the other end. So, if you got a situation where there is a lot of aggression in there, and just the right things pulling on the plug from the other end, according to the hydraulic model, you’re setting the stage for these aggressive urges to come pouring out.”
All of these utterances have to be represented by readers or listeners as part of their developing surface text, text base, situation model and context model. Im- pressionistically, this would lead to a representation of each utterance as, respec- tively, a lexico-grammatical structure, a proposition, a depiction of the referents in some state of affairs, and a representation of the position of the utterance in the on-going discourse between the sender and the addressee in some concrete situa-
tion. These postulated mental representations can be formally modeled by dis-
89
metaphorik.de 21/2011
course analysts for empirical investigation by psychologists in the following ways.
The level of surface text is meant to cater to the linguistic representation of an ut- terance, capturing word senses, syntactic structures, and so on. It shows that the first sentence has imperative mood, that there is clever ambivalent use of general and particular ‘you’ throughout the excerpt, that there is general present tense, and so on. It also shows that most of the emboldened words are used in a ‘water tank’ sense that is conventionally available from the dictionary, and presumably most people’s mental lexicon; some of the emboldened words, however, are used in their conventional metaphorical sense about the domain of the soul, also avail- able from the dictionary, such as explodes and shooting out of you, as is also ob- served by Beger (2011: 54). All of this can be explicitly and systematically mod- eled by any formal grammar that captures the structural-functional nature of sen- tences as utterances (Butler 2003).
The level of text base is intended to capture the link between words or more broadly linguistic expressions on the one hand and concepts on the other. It makes a distinction between central and more peripheral concepts by building a linearly and hierarchically ordered series of propositions that each represent the smallest idea unit that can be distinguished, in the form of a predicate and one or more arguments, all of which can be modified by further conceptual structures lower down the line. For instance, here is one possible representation of the first sentence (I am excluding this and your for the sake of convenience):
P1 (BUT S0 S1)
P2 (THINK-ABOUT ADDRESSEE TANK)
P3 (AS TANK RESERVOIR)
P4 (OF TANK WATER)
P5 (WITHIN RESERVOIR SOUL)
P6 (DRIP-INTO IMPULSES RESERVOIR)
P7 (MOD IMPULSES AGGRESSIVE)
There are various formats of propositionalization, practical application being one of the prime arguments for variation between approaches (Kintsch 1998). For
90
Steen, The value of deliberate metaphor
present purposes, these differences of detail do not affect the substance of my ar- gument. What is important to point out, though, is that surface text and text base show that, in discourse, language and thought are connected but not identical. For instance, the relative pronoun ‘that’ in the surface text is represented by the recoverable concept RESERVOIR in the text base in order to display the true con- ceptual structure of the discourse in the text base: there is a distinction between linguistic and conceptual structures.
Moving from the text base to the situation model involves the transition from ac- tivated concepts to the referents they designate in the projected text world, which is another way of looking at metaphor in thought than that which occurs for the text base. We shall later see that, precisely for metaphor, activated concepts and designated referents do not display a one-on-one relation either, just as with lin- guistic and conceptual structures in surface text and text base. In order to make this differentiation clearly visible, my impressionistic account of the first sentence and its referential representation in the situation model would be something like
this:
‘the addressee has to think about a tank of water as a reservoir within the soul, and aggressive impulses drip into that reservoir’
This may be paraphrased more felicitously and more elaborately, but that would already involve some form of enrichment. The present formulation is intended to capture the entities, relationships and attributes that must figure in the situation model if the addressee has successfully processed the first sentence. These are the elements of discourse that the addressee must attend to in working memory and these are the elements of discourse that are part of the referential representation of the first sentence.
The distinction between surface text, text base and situation model can be seen as a reflection at the level of discourse analysis of the generally accepted semiotic distinction between symbol, concept, and referent. The referential representation of discourse in a situation model has received a lot of attention in discourse psy- chology. By comparison, the context model has remained somewhat in the back- ground, its theoretical and empirical development remaining sketchy (Van Dijk
2008). Effects of constructing a context model have been concerned with, for in-
91
metaphorik.de 21/2011
stance, the difference between readers’ representations of the same text as if it were a journalistic or a literary text (Zwaan 1993), which also affects metaphor processing (Steen 1994). Different context models influence readers’ assumptions about, for instance, the goals and norms and values of the producer of a text, which in turn may interact with the ways in which the situation model is proc- essed.
I propose that the context model for our example sentence must include the fact that the sender is using a nonliteral comparison to explain a difficult concept to the addressee. This utterance is a deliberate metaphor: it explicitly instructs the addressee to think about one thing in terms of something else. The present data have the peculiarity that ‘the something else’ in fact comes first in the sentence, in the form of a tank of water, and that it is the local topic. However, overall inter- pretation of the entire excerpt suggests that it is, of course, an aspect of our men- tal behavior, namely aggression and its relation to the soul, which is depicted in terms of a water tank, not the other way around. What this example shows, there- fore, is that language users may in fact approach the target concept of a metaphor by first positioning the source concept as a distinct referent in the addressee’s mind, and then go on to use that as a base from which the target is approached.
The point here, however, is that an analytical reconstruction of the possible con- tent of the context model has to account for the communicative (or some might say ‘rhetorical’) fact that the sender instructs the addressee to construct a cross- domain mapping from a source (water in a container) to a target (emotions)—a classic conceptual metaphor in the tradition of cognitive linguistics (Kövecses
2000). The sender does so for instructive purposes that have to do with the do- main of education, which has to be contrasted with the entertaining purposes connected with the domain of literature lying behind Shakespeare’s famous first line of Sonnet 18. A tentative formulation of the context model representation of the above utterance therefore might be something like this:
‘the lecturer is asking the students to set up an explicit metaphorical com- parison between a water tank and a set of aggressive impulses to explain
the working of aggression’
92
Steen, The value of deliberate metaphor
My think aloud data for my study of understanding metaphor in literature and in journalism contained quite a few of these context constructions, about one for every five to six metaphorical expressions (Steen 1994); whether these were all and only related to deliberate metaphors I have not been able to examine yet.
In sum, following general insights in the psychology of discourse, metaphor has to be represented at various levels of discourse processing. Each of these levels captures particular aspects of the three-dimensional nature of metaphor: its lin- guistic form, its conceptual structure, including its referential structure, and its communicative function (Steen 2008; Steen 2011b). I have sketched out how this may lead to distinct representations in working memory for a deliberate meta- phor from data analyzed by Beger (2011). This is given more explicit and formal treatment in section 4, where the five-step method can be employed to examine how deliberate metaphor differs from non-deliberate metaphor and how one type of deliberate metaphor may be distinguished from another. These structural analyses hence point to different functions of metaphor in discourse and may eventually be connected to predictions about processing.
3. Analyzing metaphor in discourse: The five-step method
The original idea of the five-step method (Steen 1999a; Steen 2009) was to forge a connection between the linguistic forms of metaphor in text and talk, on the one hand, and the conceptual structures of metaphor as assumed in cognitive linguis- tics in the form of conceptual metaphors such as ARGUMENT IS WAR on the other. Since most of the work in those days did not discuss deliberate metaphor or pay much attention to such forms as simile or analogy, the role of metaphor as meta- phor in communication was not attended to as a separate issue. Typical data that needed to be accounted for included sentences like ‘Lakoff attacked Glucksberg’ or variants thereof, and the way they were used as communicative or rhetorical devices as opposed to linguistic and conceptual structures was not examined. For present purposes it is therefore most convenient to begin with an analysis of that type of metaphor, explain some of the ideas behind the five-step method, and then continue with an analysis of deliberate metaphors of the kind discussed by
Beger (2011).
93
metaphorik.de 21/2011
Table 1
Five-step analysis of ‘Lakoff attacked Glucksberg’
Steps | Analysis |
Text | Lakoff attacked Glucksberg |
1. Identification of metaphor- related words | attacked |
2. Identification of metaphor- related propositions | P1 (ATTACKs LAKOFFt GLUCKSBERGt) |
3. Identification of open meta- phorical comparison | SIM {0 F, x, y [F (LAKOFF, GLUCKSBERG)]t [ATTACK (x, y)]s} |
4. Identification of analogical structure | SIM [CRITICIZE (LAKOFF, GLUCKSBERG)]t [ATTACK (ATTACKER, ATTACKED)]s} |
5. Identification of cross-domain mapping | TARGET < SOURCE DOMAIN CRITICIZE < ATTACK LAKOFF < ATTACKER GLUCKSBERG < ATTACKED possible inferences: ARGUMENTS < WEAPONS |
Table 1 offers an overview of this five-step analysis of the metaphor ‘Lakoff at- tacked Glucksberg’. Step 1 involves the identification of any metaphor-related words in a stretch of text. In our case, the word attack is seen as a potential ex-
pression of an underlying metaphorical mapping; what the mapping is, does not
94
Steen, The value of deliberate metaphor
have to be specified for the word to be identified as metaphorical. The way in which such linguistic metaphor identification can be carried out has been thor- oughly researched, modeled and tested by the Pragglejaz Group (2007), Steen (2007) and Steen et al. (2010).
The second step displays the representation of the linguistic input in conceptual structures, as is also needed for the cognitive-linguistic approach and other ap- proaches inspired by them. As suggested above, doing so by propositionalization is regular practice in discourse psychology. There is one addition that is particu- lar to our field of study: non-metaphor related words are indexed as evoking tar- get-domain concepts whereas metaphor-related words are indexed as evoking source-domain concepts. A proposition may be said to be related to metaphor if it has at least one concept that is indexed for belonging to some source domain.
Step 3 addresses the problem that step 2 has raised: if a proposition contains a concept from another conceptual domain than the target domain, this is a poten- tial threat to the coherence of the text which needs to be resolved. This has to do with the widely noted characteristic of metaphor as a case of incongruity (Cam- eron 2003; Steen 2007). If we are dealing with metaphor we may assume that the incongruous concept comes from a source domain that can be connected to the target domain by cross-domain mapping. We can consequently postulate that, for some presupposed but unknown concepts, there is some similarity between the two incomplete propositions that are to be aligned by the suggested cross- domain mapping. This is a simple and elegant mechanism which was proposed by George Miller (1979) in an undervalued essay in Andrew Ortony’s (1979) first edition of Metaphor and Thought. Its dependence on the operation of similarity and comparison may be controversial to some, including Lakoff and Johnson (1999) as well as Glucksberg (2001), but has been supported by experimental re- search on metaphor processing by Gentner and Bowdle (2001, Genter and Bow- dle 2008; Bowdle and Gentner 2005) and theoretical work by others (Steen 2007). What I have added to Miller’s treatment is the explication that the top proposi- tion designates an incomplete state of affairs in some target domain, while the bottom proposition designates an incomplete state of affairs in some source do- main. This has been done by simple extraposition of the indices from the individ-
ual concepts to the encompassing propositional frame that includes all explicit
95
metaphorik.de 21/2011
and presupposed concepts for target and source, respectively. Note that when the open slots in the frames are filled out, in step 4, they are functionally equivalent to the source and target mental spaces used in Blending Theory (Fauconnier and Turner 1998); nothing depends on their formulation as propositions in the pre- sent approach, these are just convenient and maximally precise and explicit forms of notation (Kintsch, 1998).
As soon as the open comparison has been generated by step 3, its completion by the insertion of new concepts in each of the two frames in step 4 is a natural con- sequence. The idea is that there is an academic context in which Lakoff does something to Glucksberg which may be compared with attacking, and conven- tionally this will be interpreted as strongly criticizing; this gives us an interpreta- tion for the open slot F. This is even the conventionally metaphorical meaning of the verb to attack in the dictionary (on the systematic use of dictionaries for this analysis, see Krennmayr 2008). Conversely, the action of attacking entails two open slots a and b and these can be filled by an attacker and an attacked. Filling out the open slots in the two propositions is therefore not highly problematic, in this instance. And again, representing this analogy in a propositional format is not meant to suggest any a priori preference for propositions over other forms of representation, including, for instance, image-schemas and so on—these are em- pirical issues for behavioral research.
Step 5 then turns the analogy produced in step 4 into a differently ordered map- ping according to the format popularized by cognitive linguists. It takes each of the correspondences projected by the analogy of step 4 as separate entities and lines them up as a list of entailments. This also enables the explication of other possible entailments that are implied by the analogy, such as the presumable cor- respondence between words or arguments on the one hand and weapons or for instance fists on the other. The fact that these have not been explicitly given in the text, however, changes the status of this part of the analysis.
Table 1 labels the five steps with the appropriate cognitive-scientific terms. It uses
‘metaphor-related’ for reasons that have been explained in Steen et al. (2010) and to which we will return to below. And it shows how the most likely label of the
cross-domain mapping found in the present utterance would be CRITICISM IS
96
Steen, The value of deliberate metaphor
ATTACK. ARGUMENT IS WAR is a more general label which is clearly based on a more encompassing set of data coming from many different uses of many differ- ent terms. It raises a fundamental question about the provenance and persistence of conceptual metaphors in relation to specific instances of situated meaning (Cameron 2003; Steen 2007; Gibbs 2011c).
4. Five-steps and three dimensions
Let us now examine the connection between the five-step method demonstrated in section 3 and the three dimensions of metaphor in discourse discussed in sec- tion 2. Step 1 of the five-step method zooms in on an aspect of the representation of metaphor as part of the surface text, the fact that one of its words, attack, is used in a metaphorical as opposed to non-metaphorical sense. Step 2 involves the regular discourse-psychological transformation of linguistic materials into con- ceptual structures in the text base. Step 3 is a necessary aid to get from the con- ceptual text base, which displays the problem of conceptual incongruity, to the referential situation model, which should presumably comprise a coherent situa- tion in the target domain in which entities, relations and attributes display recog- nizably coherent properties. The target domain proposition in step 4, suggesting that Lakoff criticizes Glucksberg, is precisely this, since it designates all referents that are projected by the sentence into the text world. All of these observations have been part and parcel of the application of the five-step method in previous publications (Steen 1999a; Steen 2007; Steen 2009; Dorst 2011; Krennmayr 2011).
The three-dimensional model for metaphor now raises new questions about the alignment between the three dimensions of discourse (language, thought and communication) on the one hand and their use in the five-step method for meta- phor analysis on the other. Thus, the value of the source domain proposition in Step 4 is different for deliberate metaphor than for non deliberate metaphor, as might be the use of Step 5. Let me illustrate what may be at stake by analyzing
the deliberate water tank metaphor by means of the five-step method.
97
metaphorik.de 21/2011
4.1. Steps 1 and 2
The first step, concerning the identification of the metaphor-related words, raises a complicating detail, pertaining to the role of reservoir, which first needs to be clarified before we can proceed to the main point. It looks as if the professor in- tends reservoir to designate something that is not a concrete water tank but its emotional counterpart in the soul, something abstract rather than concrete. This also appears to be the interpretation elaborated by Beger(2011: 54).: “Our soul is conceptualized as a container that is compared to a tank of water …”. However, consultation of the Macmillan dictionary (Rundell 2002) shows that reservoir has three senses, the first two of which are concrete, while the latter is abstract but mainly literary:
1 an artificial or natural lake where water is stored so that it can be sup- plied to the houses in an area
2 a container, often part of a machine, where liquid is kept for a particular purpose
3 mainly literary a large quantity of something that can be used: +of Nearby colleges are a reservoir of talent for employers
98
Steen, The value of deliberate metaphor
Table 2
Five-step analysis of ‘think about the water tank as a reservoir within the soul …’
Steps | Analysis |
Text | But think about this tank of water as the reservoir within your soul, that aggressive impulses are dripping into |
1. Identification of metaphor-related words | About this tank of water as …reservoir within … that … drip- ping into |
2. Identification of metaphor-related propositions | P1 (BUTt S0 S1) P2 (THINK-ABOUTt ADDRESSEEt TANKs) P3 (ASt TANKs RESERVOIRs,t) P4 (OFs TANKs WATERs) P5 (WITHINs RESERVOIRs,t SOULt) P6 (DRIP-INTOs IMPULSESt RESERVOIRs,t) P7 (MODt IMPULSESt AGGRESSIVEt) |
3. Identification of open metaphorical comparison | SIM (THINK-ABOUT TARGET AS-SOURCE) {0 F, a, y [F (IMPULSES[AGGRESSIVE], ‘RESERVOIR’[WITHIN-SOUL])]t [DRIP-INTO (x, TANK-AS-RESERVOIR[WITHIN-y]])]s} |
4. Identification of analogical struc- ture | SIM (THINK-ABOUT TARGET AS-SOURCE) { [ADD-TO (IMPULSES[AGGRESSIVE], ‘RESERVOIR’[WITHIN-SOUL])]t [DRIP-INTO (WATER, TANK-AS-RESERVOIR[WITHIN-ECOSYSTEM])]s} |
5. Identification of cross-domain mapping | TARGET < SOURCE DOMAIN ADD-TO < DRIP-INTO IMPULSES[AGGRESSIVE] < WATER ‘RESERVOIR’[WITHIN-SOUL] < TANK-AS-RESERVOIR[WITHIN- ECOSYSTEM] |
On Beger’s (2011) interpretation, assuming that the professor means reservoir in its abstract sense, reservoir therefore turns out to be a metaphorically used word itself. However, its metaphorical sense comes from the same source domain as the source domain with which it is meant to be contrasted and compared, that is, the concrete sense of tank of water. This complication would involve embedding one metaphor (reservoir for large quantity) inside another one (the soul is a con- tainer) while the source domains of both metaphors are identical. This is techni-
cally possible and empirically plausible, but distracting for our present purposes.
99
metaphorik.de 21/2011
(A similar point can be made about impulse, which I am going to ignore for the sake of simplicity too.)
There is, however, another interpretation of reservoir, which is equally plausible and which would preserve its concrete meaning while simultaneously utilizing the abstract meaning. It would make reservoir ambiguous between a synonym for tank on the one hand and an abstract metaphorical interpretation or replacement of tank for some property of the soul on the other. Such double uses of words in metaphorical contexts have been labeled as ‘bridge terms’ by Kittay (1987). This interpretation can be motivated by discourse-analytical considerations in which the professor has to move from the prior introduction of the source domain ele- ment of the water tank as a local topic in its own right to its use as a means to think about the true topic of the lecture, aggression: the double, ambiguous use of reservoir as both a kind of natural container for water as well as an abstract con- tainer for emotions embodies a natural transition between the two discourse top- ics. This can be transparently shown in the five-step analysis (see Table 2). The fact that the same utterance can be given more than one interpretation is of course a given in discourse analysis and should not be seen as an issue for the five-step method, as long as it can represent both options and spell out what their implications are; since my present argument is not concerned with these issues, I will leave the comparison between the two alternative interpretations aside.
4.2. Steps 3 and 4
Assuming our current interpretation of the utterance as our starting point then, let us now have a look at some of the issues raised by deliberate metaphor. The first observation has to do with the way in which the threat to coherence in the text base by the presence of two conceptual domains is resolved in step 3 by the similarity operator. When we are dealing with a deliberate metaphor, chances are that the similarity operator is in fact explicitly expressed in the language, suggest- ing that the producer is signaling to the addressee that they have to resolve the apparent incongruity by means of a cross-domain mapping. This does not neces-
sarily have to occur, as there is a lot of deliberate metaphor that is not explicitly
100
Steen, The value of deliberate metaphor
flagged as metaphorical. However, in our example we do have such a signal or metaphor flag, think about x as y.
For deliberate metaphors like these, the assumption of a similarity operator à-la- Miller (1979) turns into more than an assumption: it reflects an explicit instruction on the part of the producer of the metaphor for the addressee to think of one thing in terms of something else. The five-step method can naturally cater to this situation by making the SIM operator more specific, in the form of adding the re- levant linguistic structure behind it as a modifier in brackets. For signaled meta- phor, the presence of the SIM operator represents a linguistic fact. For non- signaled metaphor, by contrast, the SIM operator reflects a theoretical assump- tion. It gets introduced as such in the formalism, as could be seen for ‘Lakoff at- tacked Glucksberg’ in Table 1.
The next issue raised by deliberate metaphor is the status of the source domain proposition in steps 3 and 4. In our present example, the incomplete source do- main proposition in step 3 may be held to be part of the meaning of the text as it should be represented in the working memory of the addressee. After all, it is one half of the equation which the addressee is instructed to set up as part of the local discourse. This means that the elements of the source domain which are explicitly mentioned in the text and which are represented in the incomplete bottom prop- osition in step 3, drip into, tank, reservoir, within, should all be part of the develop- ing situation model in working memory and the attention window of the ad- dressee.
For step 4, the question arises whether addressees are forced to attend to the full details of the complete source domain proposition. It is possible in theory that they just attend to the few source domain words represented in step 3 and leave it at that, concentrating instead on filling out the target domain proposition, which, after all, reflects the referential elements in the situation model that the text is about. People often make do with shallow processing (Sanford et al. 2006), and this could be one moment where that might happen. Whether shallow proc- essing takes place and decreases attention to the source domain proposition in
step 4 or not, the overall point here is that deliberate metaphor requires attention
101
metaphorik.de 21/2011
to elements of the source domain, both in steps 3 and 4, which should be visible in addressees’ construction of the situation model.
This is where the difference between deliberate and non-deliberate begins to af- fect the five-step method and reveals new opportunities for its use. We have just taken steps 3 and 4 as a representation of the situation model and argued that, for deliberate metaphor, the situation model should contain source domain elements or even a complete source domain proposition, next to the target domain propo- sition which reflects the topic of the metaphor in terms of the encompassing dis- course. To be concrete, the situation model for the water tank sentence should comprise a referential picture of the soul having some abstract reservoir in which aggressive impulses accumulate, this picture being seen as at least partly analo- gous to something dripping into a water tank which acts as a natural reservoir.
With non-deliberate metaphor, this is different. When addressees derive ‘Lakoff criticized Glucksberg’ in step 4 (assuming that this is what they do), they can rep- resent what they need. They do not need to represent the source domain ele- ments or complete proposition, ‘the attacker attacked the attackee’, as another, analogous part of their developing situation model. Indeed, it has been the claim of cognitive linguistics that people typically do not attend to the cross-domain mapping in attention; instead, they construct the mapping before then, in order to arrive at the intended target domain meaning which would be in their attention, in the form of the referential representation ‘Lakoff criticized Glucksberg’.
For non-deliberate metaphors, therefore, we might have to indicate that it is only the top proposition in step 4, the target domain proposition, which gets repre- sented in people’s situation models. The bottom proposition, representing the source domain analogue, might somehow play a part in the processes leading up to the product of comprehension in working memory, but presumably does not end up there itself. It should not be part of the situation model for non-deliberate metaphors. For deliberate metaphors, the source domain proposition should at least partly be included in the situation model. These are testable claims which
we are currently undertaking to investigate in our lab.
102
Steen, The value of deliberate metaphor
4.3. Steps 4 and 5
If this argument is acceptable as a plausible account of how metaphor gets repre- sented in working memory and the elements of the situation model that are at- tended to, and if step 4 of the five-step method is accepted as one useful way of representing these mental products, with possibly different outcomes for non- deliberate versus deliberate metaphors, then this may also form a basis for re- thinking step 5. The original aim of step 5 was to capture the full cross-domain mapping that might be related to the local analogy derived in step 4, so that ‘La- koff attacked Glucksberg’ could be connected in some controllable fashion to ARGUMENT IS WAR, and the tank of water metaphor could be connected to EMOTIONS ARE LIQUIDS IN A CONTAINER. The information in step 5 might be useful for constructing links with postulated conceptual metaphors in order to test how easy and reliable such connections might be formulated. Implicit in such an un- dertaking has been the idea that such elaborations would also throw light on the role of the conceptual metaphor in communication. And this is still one way in which any analysis of metaphor in discourse could proceed after step 4 has been completed.
However, I now feel that a more interesting use of step 5 would be to see it as representing the communicative dimension of metaphor, which would be useful as input for the ongoing construction of a context model for the discourse as a whole. As noted above, an impressionistic account of that representation might look like this:
‘the lecturer is asking the students to set up an explicit metaphorical com- parison between a water tank and a set of aggressive impulses to explain the working of aggression’
It is now clear that such a representation can be derived from the preceding in- formation in step 4, with additional default information added from the context of the genre event the utterance is part of. The crucial point, however, is that it is the presence of an alien domain in the situation model (the source domain propo- sitions in step 3 and 4) that may be conditio sine qua non for such a context model
to be enabled. If we compare this with ‘Lakoff attacked Glucksberg’, the context
103
metaphorik.de 21/2011
model might read something like ‘the sender is informing the addressee that La- koff criticized Glucksberg in order to add this event to a developing account’, for instance one in which the sender himself is either praising or criticizing Lakoff.
In other words, I am suggesting that there may be a close and differential connec- tion between the output of step 4 for non-deliberate versus deliberate metaphors on the one hand and the content of the context model on the other hand: if ele- ments of the source domain are represented as such and hence presumably at- tended to in the situation model, as is the case with deliberate metaphor, then they constitute a case of discourse incongruity which can be represented as such in the context model as well; if elements of the source domain are not represented as such and hence presumably not attended to in the situation model, as is the case with non-deliberate metaphor, then they do not constitute a case of dis- course incongruity and do not get represented as such in the context model ei- ther. The function of such an analysis would be to explain why some metaphors raise questions as to their use and point, whereas others do not. If the context model displays a representation like the one for the water tank above, this affords a platform for wondering about the rhetorical aim of the sender as well as the means and success of his action. If the context model only say something like ‘the sender is informing the addressee that Lakoff criticized Glucksberg in order to add this event to a developing account’, then there is no ground in the context model for continuing with paying any sort of attention to what technically and perhaps subconsciously is a metaphor. Again, my think aloud data for under- standing metaphor in literary versus journalistic text processing (Steen 1994) clearly showed signs of the possibility of these processes, but in just one out of about five to six cases of metaphor—these may have been the deliberate ones, and we are now preparing to address this issue head-on.
In sum, the five-step method may be closely aligned with the various mental rep- resentations that people are assumed to make during discourse processing. The discovery of deliberate metaphor as an important manifestation of the communi- cative dimension of metaphor in discourse pointed out that the context model might be more important for the five step method than originally realized. In- deed, it now looks as if the content of the context model may be partly dependent
on the representation of metaphor in step 4, which captures the situation model.
104
Steen, The value of deliberate metaphor
This became particularly clear when we looked at the difference between non- deliberate versus deliberate metaphor, which revealed that the crucial issue there may have to do with the presence or absence of the source domain in the situa- tion model (step 4). If the source domain is part of what is attended to in the situ- ation model, as is the case for deliberate metaphor, then it may give rise to a con- text model for metaphor that pays attention to metaphor as metaphor, since there is some form of discourse incongruity in the situation model; if the source do- main is not part of the situation model, as is the case for non-deliberate metaphor, then there is no ground for this to occur.
5. Conclusion: The value of deliberate metaphor
The crux of this proposal lies in the presence of the source domain in attention, in the situation model. How can we know that this is or is not the case? How can we decide whether to model the representation of a metaphor in discourse process- ing as comprising just the target domain proposition in step 4, or whether it also includes the role of some source domain elements or even a complete source do- main proposition? My answer over the past few years has been to decide whether the metaphor is used deliberately or not. However, that is a controversial matter and even though there are many clear cases there probably are just as many un- clear ones. Krennmayr (2011) has formulated some first guidelines but we have now started a new project with Gudrun Reijnierse that is fully devoted to this is- sue.
However, there are other solutions. I have two suggestions. First, we might want to turn the situation with deliberateness and representation of the source domain in attention round, and treat the whole matter as an empirical question. Since the formalism allows for source domain information to play a role until at least step
4, behavioral research could attempt to test under which conditions source do- main activation remains present in the situation model. I would predict that this would be the case for novel metaphors, difficult metaphors, metaphors that are signaled, metaphors that are extended or part of a conspicuous series of meta- phorical expressions, and metaphors that are combined with other figures of
speech, such as hyperbole or irony. Many of these might also be experienced as
105
metaphorik.de 21/2011
deliberate, but we might even want to treat this as a post-hoc effect on the part of the recipient, which would make this proposal compatible with some of the ar- guments advanced by Gibbs (2011a; Gibbs 2011 b). How all of this might work for production would be a rather different affair, and how this would be affected by using different media of communication (speech versus writing) would have to be examined as well. These are challenges that are more general for all meta- phor research.
The second suggestion would be to take more seriously the possibility that many metaphors are so conventionalized that they may not give rise to activation of source domain concepts or cross-domain mappings. Thus, even though step 1 would show that attack in ‘Lakoff attacked Glucksberg’ is linguistically meta- phorical (Steen et al. 2010), this might still be argued to lead to fast lexical disam- biguation and immediate activation of the appropriate target domain concept ATTACK in the sense of ‘criticize’ by the addressee. If that were to happen, the text base would not show any source domain concepts that would require incongru- ity resolution by triggering step 3, for it would simply say ATTACK* LAKOFF GLUCKSBERG, the star indicating that we are dealing with the metaphorical con- cept ‘attack’ (Glucksberg and Haught 2006). Instead, we would immediately pro- ceed to step 4, which would only have a target domain representation of the sen- tence, in the form of ‘Lakoff attacked* Glucksberg’, and there would be no issue about the presence or absence of source domain concepts and referents.
Even if lexical disambiguation would allow for the activation of both concrete and abstract versions of the concept ATTACK in the text base, this might still allow for a process of suppression of the irrelevant source domain concepts that is fast enough to prevent the start of any sort of mapping. The potential conceptual in- congruity between source and target domain concepts might then be gone before it can trigger a resolution process. This could happen with most conventional me- taphors that are not revitalized as metaphors in deliberate use (Giora 2003, Giora
2008).
Note that this implies that the representation of ATTACK as a source domain con- cept in Table 1 above represents just one possibility. It was chosen in order to re-
flect the assumptions of cognitive linguistics, in order to spell out the conse-
106
Steen, The value of deliberate metaphor
quences of such an assumption in the context of discourse analysis. However, this does not mean that other models of lexical and conceptual processing are not equally interesting and merit formal as well behavioral examination as well.
Both suggestions imply a need for precise formal elaboration and experimental testing of the assumptions that metaphor scholars bring to bear on their view of how metaphor is understood. Deliberate metaphor clearly offers new perspec- tives on this research area. Its analysis by the slightly adjusted five-step method raises some very concrete questions for psycholinguistic research, and not just for deliberate metaphor, but for the whole gamut of metaphor use in discourse. That is the value of deliberate metaphor.
References
Beger, Anke (2011): “Deliberate metaphors? An exploration of the choice and functions of metaphors in US-American college lectures”, in: metaphorik.de
20, 39-60.
Bowdle, Brian F., & Gentner, Dedre (2005): “The career of metaphor”, in: Psycho- logical Review, 112(1), 193-216.
Butler, Chris (2003): Structure and function: A guide to three major structural- functional theories. Amsterdam: Benjamins.
Cameron, Lynne (2003): Metaphor in educational discourse. London and New York: Continuum.
Charteris-Black, Jonathan (2004): Corpus approaches to critical metaphor analysis.
London: Palgrave MacMillan.
Charteris-Black, Jonathan/ Musolff, Andreas (2003): “'Battered Hero' or 'Innocent Victim'? A comparative study of metaphors for euro trading in British and German financial reporting”, in: English for Specific Purposes, 22, 153-176.
Deignan, Alice (2011): “Deliberateness is not unique to metaphor: A response to
Gibbs”, in: Metaphor and the Social World 1(1), 67-69.
Dorst, Aletta G. (2011): “Personification in discourse: linguistic forms, conceptual structures and communicative functions”, in: Language and Literature 20(2),
113-135.
Fauconnier, Gilles/Turner, Mark (1998): “Conceptual integration networks”, in:
Cognitive Science 22, 133-187.
107
metaphorik.de 21/2011
Gentner, Dedre/Bowdle, Brian F. (2001): “Convention, form, and figurative lan- guage processing”, in: Metaphor and Symbol, 16(3 & 4), 223-248.
Gentner, Dedre/Bowdle, Brian F. (2008): “Metaphor as structure-mapping”, in: Gibbs, Raymond W. (Ed.): The Cambridge Handbook of metaphor and Thought, Cambridge etc: Cambridge University Press;109-128.
Gibbs, Raymond W. (2011a): “Are ‘deliberate’ metaphors really deliberate? A question of human consciousness and action”, in: Metaphor and the Social World 1(1), 26-52.
Gibbs, Raymond W. (2011b): “Advancing the debate on deliberate metaphor”, in:
Metaphor and the Social World 1(1), 67-69.
Gibbs, Raymond W. (2011c): “Evaluating conceptual metaphor theory”, in: Dis- course Processes 48 (8), 529-562.
Giora, Rachel (2003): On our mind: Salience, context, and figurative language. New
York: Oxford University Press.
Giora, Rachel (2008): “Is metaphor unique?”, in: Gibbs, Raymond W. (Ed.): The Cambridge handbook of metaphor and thought, Cambridge etc: Cambridge Uni- versity Press 143-160.
Glucksberg, Sam (2001): Understanding figurative language: From metaphors to idi- oms. Oxford and New York: Oxford University Press.
Glucksberg, Sam/Haught, Catrinel (2006): “On the relation between metaphor and simile: When comparison fails”, in: Mind & Language, 21, 360-378.
Graesser, Art C./Millis, Keith (2011): “Discourse and cognition”, in: Van Dijk, Teun (Ed), Discourse Studies: A Multidisciplinary Introduction, London: Sage,
126-142.
Graesser, Art C./Mills, Keith K./ Zwaan, Rolf A. (1997): „Discourse comprehen- sion“, in: Annual Review of Psychology, 48, 163–189.
Kintsch, Walter (1998): Comprehension: A paradigm for cognition. Cambridge: Cam- bridge University Press.
Kittay, Eva Feder (1987): Metaphor: Its cognitive force and linguistic structure. Ox- ford: Oxford University Press.
Kövecses, Zoltán (2000): Metaphor and emotion: Language, culture, and body in hu- man feeling. Cambridge: Cambridge University Press.
Krennmayr, T. (2008). Using Dictionaries in Linguistic Metaphor Identification. In
N.L. Johannesson & D.C. Minugh (eds.), Selected Papers from the 2006 and
108
Steen, The value of deliberate metaphor
2007 Stockholm Metaphor Festivals. Stockholm: Department of English, Stock- holm University, 97-115.
Krennmayr, Tina (2011): Metaphor in newspapers. LOT Dissertation Series, 276.
Utrecht: LOT
Lakoff, George/Johnson, Mark (1999): Philosophy in the flesh: The embodied mind and its challenge to western thought. New York: Basic Books.
Macnamara, Danielle S./Magliano, Joe (2009): “Toward a comprehensive model of comprehension”, in: Ross, Brian (ed.): The Psychology of Learning and Moti- vation, Vol. 51,. Burlington: Academic Press, 297-384.
Miller, George A. (1979): “Images and models, similes and metaphors”, in: Or- tony, Andrew (ed.): Metaphor and thought, Cambridge: Cambridge University Press, 357-400.
Müller, Cornelia (2011): “Are ‘deliberate’ metaphors really special? Deliberate- ness in the light of metaphor activation”, in: Metaphor and the Social World
1(1), 67-69.
Ortony, A. (Ed.) (1979): Metaphor and thought. Cambridge: Cambridge University
Press.
Perfetti, Charles A. (1999): “Comprehending written language: A blueprint of the reader”, in: Brown, C./Hagoort, P. (Eds.). The neurocognition of language. Ox- ford: Oxford University Press, 167-208.
Pragglejaz Group (2007): “MIP: A method for identifying metaphorically used words in discourse”, in: Metaphor and Symbol, 22(1), 1-39.
Rundell, Michael (2002): Macmillan English Dictionary for Advanced Learners. Ox- ford: Macmillan.
Sanford, A.J.S/Sanford, Anthony J./ Molle, J.,/Emmott, Catherine (2006): “Shal- low processing and attention capture in written and spoken discourse”, in: Discourse Processes 42, 109-130.
Semino, E. (2008): Metaphor in discourse. Cambridge: Cambridge University Press. Shen, Yeshayahu/Balaban, N. (1999): “Metaphorical (in)coherence in discourse”,
in: Discourse Processes, 28, 139-154.
Steen, Gerard J. (1994): Understanding metaphor in literature: An empirical approach.
London: Longman.
Steen, Gerard J. (1999a): “From linguistic to conceptual metaphor in five steps”, in: Gibbs, Raymond/Steen, Geerard (eds.): Metaphor in cognitive linguistics Amsterdam: John Benjamins, 57-77.
109
metaphorik.de 21/2011
Steen, Gerard J. (1999b): “Metaphor and discourse: towards a linguistic checklist for metaphor analysis”, in: Cameron, Lynne/Low, G (eds.): Researching and applying metaphor , Cambridge: Cambridge University Press, 81-104.
Steen, Gerard J. (2004): “Can discourse properties of metaphor affect metaphor recognition?”, in: Journal of Pragmatics, 36(7), 1295-1313.
Steen, G.J. (2006): “Discourse functions of metaphor: An experiment in affect”, in: Benczes, R. & Csabi, S. (eds.): The metaphors of sixty: Papers presented on the oc- casion of the 60th birthday of Zoltan Kövecses, Budapest, Hungary: School of English and American Studies, Eotvos Lorand University, 236-244.
Steen, Gerard J. (2007): Finding metaphor in grammar and usage: A methodological analysis of theory and research. Amsterdam: John Benjamins.
Steen, Gerard J. (2008): “The paradox of metaphor: Why we need a three- dimensional model of metaphor”, in: Metaphor and Symbol, 23(4), 213-241.
Steen, Gerard J. (2009): “From linguistic form to conceptual structure in five steps: Analyzing metaphor in poetry”, in: Brône, G./Vandaele, J. (eds.), Cognitive poetics: Goals, gains, gaps, Berlin and New York: Mouton de Gruyter, 197-226.
Steen, Gerard/Dorst, Aletta/Herrmann, Berenike/Kaal, Anna/Krennmayr, Ti- na/Pasma, Tryntje (2010a): A Method for Linguistic Metaphor Identification: From MIP to MIPVU, Amsterdam/Philadelphia.
Steen, Gerard J. (2011a): “What does 'really deliberate' really mean? More though- ts on metaphor and consciousness”, in: Metaphor and the Social World 1(1), 53-
56.
Steen, Gerard J. (2011b): “The contemporary theory of metaphor - now new and improved!”, in: Review of Cognitive Linguistics 9(1), 26-64.
Steen, Gerard J. (in press): “Deliberate metaphor affords conscious metaphorical cognition”, in: Cognitive Semiotics.
Van Dijk, Teun A./Kintsch, Walter (1983): Strategies of discourse comprehension.
New York: Academic Press.
Van Dijk, Teun A. (2008): Discourse and Context: A sociocognitive approach. Cam- bridge: Cambridge University Press.
Zwaan, Rolf A. (1993): Aspects of Literary Comprehension: A cognitive approach. Am- sterdam: Benjamins.
110