A cognitive semiotic exploration of metaphors in Greek street art

2.1 CMT and recent developments

With the publication of Lakoff and Johnson’s Metaphors We Live By (1980) and the rise of studies on metaphor in cognitive linguistics (e.g., Gibbs 1994; Grady 1997; Johnson 1987, Johnson 2007; Kövecses 2005, Kövecses 2010a; Lakoff and Johnson 1999, Lakoff and Johnson 2003; Lakoff and Turner 1989; Müller 2008a), the focus of investigations started to shift from the analysis of creative and deliberate expressions onto everyday expressions like we are stuck, he is over the hill, and the hypothetical mental processes underlying them. It was claimed that “our ordinary conceptual system, in terms of which we both think and act, is fundamentally metaphorical in nature” (Lakoff and Johnson 1980: 3). Thus, the traditional notion of metaphor as a poetic and rhetorical device was rejected, and metaphor was redefined as a conceptual scheme of cognitive correspondences (Lakoff and Johnson 1980, Lakoff and Johnson 1999, Lakoff and Johnson 2003) between a source domain and a target domain, as shown in diagrams such as Table 1, and referred to using the TARGET IS SOURCE schema (Lakoff and Johnson 1980).^[3]

In the late 1990s, the CMT approach was elaborated by Grady (1997) who distinguished between primary and complex (compound) metaphors. Hampe (2017) points out that primary metaphors may be seen as deeply embodied cognitive analogies, which arise from pan-human bodily sensory-motor experiences. Several scholars have highlighted that in this sense many conceptual metaphors (such as MORE IS UP, GOOD IS UP) discussed in the literature are primary metaphors and potentially universal since they correspond to certain bodily experiences that are shared by all human beings, as opposed to the more complex (compound) metaphors like LIFE IS GAMBLING GAME or LIFE IS JOURNEY (Grady 1997: 288; Grady and Ascoli 2017; Kövecses 2009, Kövecses 2010b). According to the theory, primary metaphors are universal, automatic, and outside of consciousness:

We do not have a choice as to whether to acquire and use primary metaphor. Just by functioning normally in the world, we automatically and unconsciously acquire and use a vast number of such metaphors […] They are a consequence of the nature of our brains, our bodies, and the world we inhabit. (Lakoff and Johnson 1999: 59)

While influential, the theoretical framework of CMT has often been criticized by a number of researchers, who represent distinct approaches in cognitive linguistics and other neighbouring disciplines like cognitive psychology, media and communication, anthropology, and literature studies. The following issues have been the focus of the debate.

Universal or cultural? As pointed out, CMT has emphasized that metaphor is grounded in pan-human bodily experiences, paying special attention to universal mappings between source and target domains (Lakoff and Johnson 1999: 47, 56–57). This means that all conceptual metaphors, according to CMT, are essentially independent of any specific language and primary metaphors in particular are independent of culture and background knowledge. On the other hand, it has been pointed out that even primary metaphors depend on socio-cultural and historical contexts (Forceville 2017; Kövecses 2010a; Kövecses 2017). Casasanto (2017) presents a review of studies, demonstrating cross-linguistic, cross-cultural, and cross-individual variation. This would seem to be in favour of metaphor theories that highlight creativity and dynamicity in both creation and interpretation (e.g., Cameron et al. 2009; Müller 2008a).

Unconscious or conscious? CMT not only emphasized universality, but also the claim that mappings such as that in Table 1 (aka conceptual metaphors), are essentially part of the “cognitive unconscious” (Lakoff and Johnson 1999). Zlatev (2011) argues that that notion is problematic philosophically and empirically, especially given analyses (and most examples) of metaphors, which are based on the researchers’ intuitions (cf. Itkonen 2008b). Furthermore, over the past decade, a number of cognitive linguists and discourse analysts have critically addressed this issue and drawn attention to the role of deliberate (intended) metaphor use in language and communication (e.g., Cameron 2003; Charteris-Black 2004; Müller 2008a; Semino 2008; Steen 2008, Steen 2009, Steen 2011, Steen 2017; Steen et al. 2010).

Methodology: metaphors in context? As pointed above, the original methodology of producing metaphors and analysing them on the basis of researcher-intuition was subsequently extended with experimentation (Boroditsky 2000; Casasanto 2017; Gibbs 2017a, Gibbs 2017b). In addition, many have argued that this should be complemented with a discursive approach to metaphor (Cameron 2003, Cameron 2016; Charteris-Black 2004; Müller 2008a; Musolff 2004; Semino and Demjén 2017; Steen et al. 2010; Zinken 2007; Zinken et al. 2008; Zinken and Musolff 2009) with the help of corpus studies as well as detailed studies of discourse. For example, Semino (2008) has argued that decontextualized cognitive analogies in the traditional X IS Y form (and their underlying mappings) cannot fully account for metaphors in real-life social discourse. Several analysts have advocated the use of the multiple “spaces” of conceptual-integration (aka blending) theory as allowing more flexibility for such analyses than CMT (e.g., Fauconnier and Turner 2002; Dancygier and Sweetser 2014; Kövecses 2017).

In language or beyond? Even if CMT claims that metaphor is not a linguistic but cognitive phenomenon, most of the evidence for the theory comes from language, and only to some degree from other semiotic systems (with new developments) like gesture (e.g., Cienki 2013, Cienki in press; Mittelberg 2008; Müller 2008b) and depiction (e.g., Bolognesi et al. 2018; Forceville 2009). If metaphor is indeed not limited to language, this would lead us to inquire into how semiotic systems differ in the way metaphors can be expressed and how they interact with one another. Research along these lines has been accumulating on a number of media and genres with specific focus on the semiotic system of depiction realized in genres such as print advertising (e.g., Forceville 2017; Pérez-Sobrino 2017; Bolognesi and Strik Lievers 2019), film (e.g., Forceville 2016; Coëgnarts and Kravanja 2015; Fahlenbrach 2016), and cartoons (e.g., El Refaie 2009; Teng 2009). This is usually done under the heading of “multimodality” but, as we propose in Section 3, this is a rather ambiguous term.

To summarize, it has become clear over the course of these debates that culture, consciousness and context are crucial to both production (creation) and comprehension (interpretation) of metaphors (Musolff 2004; Charteris-Black 2004; Kövecses 2005, Kövecses 2015; Pragglejaz Group 2007; Semino 2008; Zinken and Musolff 2009). In other words, there is wide agreement among many metaphor scholars that at least the original CMT has failed to acknowledge the essence of metaphorical creativity, and new theories have been proposed such as discourse metaphor theory (Zinken 2007) and deliberate metaphor theory (Steen 2011). However, such have mostly focused on metaphor in language.^[4] The challenge is to go beyond language, and for ideas on how to do so, it is natural to go to semiotics.

2.2 Semiotic approaches to metaphor

Semiotic theories of metaphor differ enormously but meet in at least one respect: the assumption that metaphors are a particular kind of sign (process). Unfortunately, this does not amount to much common ground, as sign definitions differ extensively among semioticians. These fall very roughly in two different traditions: a Saussurian structuralist tradition that emphasizes convention and “arbitrariness,” and a Peircean tradition that highlights interpretation (cf. Chandlers 2007). In a kind of synthesis of these traditions, as well as ideas from Husserlian phenomenology, Sonesson (2010, 2014, and 2015) has developed a definition of the sign as a kind of meaning making process that presupposes conscious awareness and requires the experiencing subject to both associate and differentiate expression and content. This gives a sign concept that is quite intuitive and consistent with common sense, as it renders (most) words, representational drawings, many gestures, and natural symptoms like smoke as signs — provided, of course, the required interpretative abilities of a subject to understand them as such. On the other hand, simple stimulus-response associations, or acts of direct perception are certainly meaningful, but not sign-mediated processes. In addition, as we discuss in Section 3, signs typically form systems (like language, gesture, and depiction), with more or less complex inter-sign relations (Zlatev 2019).

From this perspective, the well-known sign-typology based on the work of Peirce: iconic signs (icons), indexical signs (indices), and symbolic signs (symbols) can be understood as derived from the predominant semiotic ground that links expression and content during interpretation, where there are nearly always several grounds combined in a single sign (Jakobson 1965). The three types of grounds are iconicity (similarity), indexicality (contiguity or part-whole), and symbolicity (conventionality). Thus, icons and indices are motivated while symbolic signs are conventional although not arbitrary, as they often involve the other two grounds as well.^[5] All these kinds of signs are involved in acts of communication such as those expressed by the street artworks that we analyse and review in Section 4, where we show that the semiotic grounds of indexicality and symbolicity (based on socio-historical background knowledge and context) interact closely with iconicity. But what makes a sign process metaphorical?

In the semiotic theory of Peirce (1931/1974), there are three types of (hypo)icons: imagistic (images), diagrammatic (diagrams), and metaphorical (metaphors), all grounded in similarity. In the case of the image, expression and content share simple qualities, such as the colour samples listed in a colour palette (Sonesson 2014). In the case of diagrams, it is a matter of similarity of relations rather than properties (Lenninger 2012). Thus, (Peirce 1931/1974: 2.277) defines diagrammatic iconic signs as “those, which represent the relations […] of the parts of one thing by analogous relations in their own parts.” The definition of metaphor given by Peirce is famously unclear,^[6] but is generally taken to be “one level higher” than that of diagram (Hausman et al. 1996; Lance et al. 1996), implying that the diagram/analogy/mapping is waiting to be discovered and creatively interpreted in various possible and contextually appropriate ways (Sonesson 2015). Thus, the static cross-domain mapping along the lines of CMT’s LIFE IS GAMBLING GAME (see Table 1) is in itself a diagram, while a (somewhat) creative expression like (1), taken from a popular ABBA song, can function as a metaphor that requires the kind of diagram shown in Table 1 for its interpretation.

This analysis conforms with the view that metaphors are the most complex iconic signs (Peirce 1931/1974) especially when understood as creative, emergent, and dynamic processes, socio-culturally grounded and contextually influenced: a view that has been gaining currency in both cognitive linguistics and semiotics (Kövecses, 2015; Müller, 2008a; Sonesson, 2015). As shown in Figure 2, however, metaphors are only one kind of sign (process) and can be understood only in relation to and interaction with other kinds of signs. This implies that the (verbo-) pictorial metaphors in street art that we analyse in Section 4 may interact with other rhetorical figures that are supported by relations of contiguity (metonymies) and part-whole relations (synecdoches)^[7], as seen in Figure 2 below:

Figure 2:

A hierarchical organization of sign relations.

Turning from figurativity in the semiotic system of language to that of depiction raises additional issues. Ever since Barthes’s (1964) classical paper “La rhétorique de l’image,” where he proposed that the tomato in the advertisement served as a metaphor for Italy, French structuralists have worked on different taxonomies of rhetorical figures in images comparable to those elaborated for language since classical antiquity (Durand 1983). One of the foremost of these was developed by the group of researchers known as Groupe µ (1970, 1976). In their later work, however, Groupe µ (1992) rejected the typology of rhetorical figures inherited from verbal rhetoric, and instead proposed a cross-classification of elements which conflict with everyday expectations, creating a degree of incongruity. This latter notion figures in a number of definitions of metaphor, such as in the “interaction theory” of Black (1979), where a (good) metaphor is supposed to offer unexpected ways of thinking, as it brings together two concepts (tenor/target and vehicle/source) that persist in a state of tension. Sonesson (1996, 1997, 2010, 2014, and 2015) has developed these ideas in relation to the phenomenology of perception, according to which our experiences are based on expectancies that may be confirmed or disappointed. Sonesson’s analysis of figuration in images focuses on part-whole relations (and thus indexicality), proposing that most disappointed expectancies involve both the presence of something unexpected and the absence of something expected, and thus incongruity. This can be seen in Figure 3, where the Greek flag on the roll of toilet paper is certainly unexpected, highlighting both similarities (in physical form) and dissimilarities (in function), as discussed in detail in Stampoulidis and Bolognesi under review.

We should note here a difference between the semiotic systems of language and that of depiction. In language, metaphoric expressions typically manifest themselves with an explicit directionality, conveyed by the order of the constituents and syntax in general: both in so-called direct metaphors (e.g., “my job is a jail”) and in indirect metaphors (e.g., “she devoured the novel”) (see Steen 2017) it is clear what constitutes the source and what constitutes the target of the metaphor. In the semiotic system of depiction such directionality is less clearly expressed. Anticipating the analysis presented in Section 4, on the level of metaphoric expression the directionality typically remains unclear, until the viewers integrate what they see with genre-related knowledge, conventions, the nature of the depicting surface, and other types of contextual information. Once such knowledge is integrated, it becomes clearer which part of the pictorial representation functions as source and which one functions as target, and the directionality clearly emerges, to allow viewers to interpret the metaphor on the basis of iconicity. The presence of pictorial incongruity (a strong form of tension) and iconicity between source and target concepts suggest that a figurative interpretation is needed, but only through the integration of socio-cultural and context-specific knowledge (indexical and conventional world knowledge) are the viewers able to make connections with their experiential lifeworld, and interpret the metaphor as such (see Stampoulidis and Bolognesi under review).

Figure 3:

A unisemiotic monomodal street artwork by an unknown creator. Metaphor: GREEK FLAG ≡ TOILET PAPER. Photograph by Stampoulidis, July 2015.

Thus, Sonesson (2015) proposes that metaphor creation and interpretation is a kind of discovery procedure, since it is possible to find similarities about two things involved in the metaphor for which you were not aware of beforehand and draw open-ended inferences (see Section 4). This is a productive idea, but we find that requiring the property of incongruity for metaphoricity is too restrictive, in particular for moderately creative (verbal) metaphors, such as that in (1). In a way, this approach goes to the opposite extreme of CMT, acknowledging as metaphorical only fully creative and non-conventional (iconic) processes, while metaphors in both language and depiction can be more or less conventionalized (sedimented) in the cultural knowledge of a given society (e.g., Svanlund 2007), as we explicate in section 3.2.^[8]

2.3 Summary

As reviewed in this section, cognitive linguistic and semiotic approaches to metaphors have often grappled with similar problems of analysing and even defining metaphors in relation to other figurative processes, allowing some account for the continuum of conventionality and/or creativity. At the same time, there is considerable terminological ambiguity and a suspicion bordering on hostility between the two traditions. With the framework that we present in the following section and apply in a study of Greek street art images in Section 4, we hope to be able to provide a possible common ground.

3.1 Methodology and central concepts

As stated in the introduction, cognitive semiotics combines concepts and methods from cognitive linguistics, semiotics, and cognitive science with the aim to study the multi-faceted notion of meaning as comprehensively as possible. One of its core principles, in the spirit of phenomenology and hermeneutics, is the conceptual-empirical loop mentioned in Section 1, which requires that we start with few preconceptions and to formulate definitions of the objects of study that are as intuitive as possible. Using these definitions, we perform empirical analyses using observation and experimentation in order to return to the concepts we started with and possibly improve them. A second principle consistent with the first is methodological triangulation, combining the use of first-, second-, and third-person methods in the study of specific semiotic phenomena based on the perspective of the researcher, like intuition, empathy, and experimentation, respectively (Zlatev 2009, 2015). These principles are exemplified in Section 4 in reviewing the study of Stampoulidis and Bolognesi under review.

The concept of sign is central for cognitive semiotics, but not as co-extensional with meaning in general, as in many schools of semiotics (Chandler 2007). Rather, it is a special kind of process that presupposes the conscious awareness on the side of the interpreter that a given expression, in one or another material medium, represents a particular content in some respects, from a certain perspective, consistent with the theory espoused by Sonesson (e.g., 2010) and summarized at the beginning of Section 2.2.

Signs form systems with more or less complex inter-sign relations and properties that are in part determined by the medium employed by the particular system. Language — in either the spoken (speech), bodily (signed languages), or written (text) medium — is the best-known semiotic system, allowing the expression of an endless possibility of different meanings. This is largely due to the relations between its signs (paradigmatic, syntagmatic, hierarchical), which correspond to what is generally known as “grammar” (see Zlatev 2019). Speech and writing are sub-systems of language with properties such as “rapid fading” (Hockett 1960) differentiating them (see Table 1). Another semiotic system is that of gestures, understood as intended expressive bodily movements that are visually perceivable (Kendon 2004). Most relevant for present purposes, however, is the semiotic system of depiction, realized in various media from sand drawings and wall paintings to the numerous technologies that all involve the production of lines and patches of contrasting colours on surfaces in such a way so as to create a “likeness” to real or imaginary objects or events (Green 2014). Table 1 shows some of the properties of these three semiotic systems, with “material” referring to extra-bodily resources like a surface affording inscription, and “double articulation” to the presence of a level of systematic elements like phonemes that are meaningless in themselves. “Syntagmatic relations” are linear combinations of signs and the notion of semiotic ground was discussed in Section 2.2.

While each one of these systems may be used independently, most spontaneous human communication involves the combination of two or more, e.g., polysemiotic communication (Zlatev 2019). This allows complex interactions of sign use, where the different expressive potentials of the systems interplay with and balance one another in ways that remain to be explored in detail.

Some research in this direction is performed under the heading of “multimodality” (Jewitt et al. 2016), in several different research traditions, with rather conflicting terminology depending on the context and the goals of the researcher. In cognitive linguistics (for a review, see Devylder in press), the term “modality” is often used to refer more or less to what we mean by semiotic system, for example, by calling verbo-pictorial images such as that in Figure 8 “multimodal metaphors” (e.g., Forceville 2017). In gesture studies, speech and gestures are commonly referred to as “communicative modalities,” and language itself is often considered “multimodal” (Vigliocco et al. 2014). On the other hand, in the tradition of social semiotics (e.g., Bateman 2011; Kress 2009; van Leeuwen 2005), one considers the combination of “modes” such as speech, text, image (still and moving), colour, music, typography, font, layout, design, and so on under the notion of “multimodality.”^[9] However, as pointed out by Green (2014: 9–10): “there is no theoretical limit to the number of modes that may be recognized in various socio-cultural contexts, and this leads to an abundance of modes that are difficult to compare,” especially with research outside of social semiotics. Finally, and what is arguably the least controversial use of the term, in most of psychology “modality” is used to refer to the different senses: sight (visual), hearing (auditory), smell (olfactory), touch (tactile), taste (gustatory), and possibly others like proprioception (body awareness). Perception as a whole can be said to be multimodal (e.g., Fulkerson 2014; O’Callaghan 2012).

Given this extensive ambiguity of the term “multimodality” (see also Adami 2016; Green 2014; Devylder in press), we constrain the use of the term to the synergy of two or more different sensory modalities in the act of perception (see Table 2). A street artwork may thus be either unimodal, if perceived only visually, or multimodal, if also perceived through other sensory modalities, such as tactile (if we touch it while walking down the streets). As pointed out earlier, we use the term polysemiotic communication for the synergy of two or more semiotic systems (see Table 2). For example, a work of street art consisting of verbal text (language) and pictorial elements (depiction) is clearly a form of polysemiotic communication, instantiated in the particular socio-cultural medium of street art. Conversely, the street artwork depicted in Figure 4 might be considered as an example of unisemiotic communication (only the semiotic system of depiction is present) but is still (potentially) multimodal since it may trigger multiple senses in the viewer, such as sight and touch.^[10] The street artwork shown in Figure 5, instead, is polysemiotic and (potentially) multimodal, because it exploits features (like the texture of the pig) related to the sense of touch, besides features (like shape and colour) related to the sense of sight.^[11] Metaphorical images such as that shown in Figure 3 should be regarded as unisemiotic and monomodal, while those that combine language and depiction as that in Figure 7 (Section 4) are polysemiotic and monomodal.

Figure 4:

A unisemiotic and (potentially) multimodal street artwork by Bleeps.gr. Photo courtesy of the artist.

Figure 5:

A polysemiotic and (potentially) multimodal street artwork by an anonymous queer group: Stop Homo-Trans-Phobia. Photograph by Ilaria Hoppe, September 2010, Berlin (Kreuzberg), Germany.

3.2 The motivation and sedimentation model

Inspired by concepts from phenomenology (Husserl 1901/1970; Merleau-Ponty Merleau-Ponty 1945/1962) and integral linguistics (Coseriu 1985, 2000), the Motivation and Sedimentation Model (MSM) distinguishes between three basic levels of meaning (the embodied, the sedimented, and the situated), and links these with two operations: motivation and sedimentation.^[12] It has been applied to the understanding of language norms (Zlatev and Blomberg in press), to the debate on linguistic relativity (Blomberg and Zlatev in press), and most recently to metaphor (Devylder and Zlatev in press). It may be illustrated as in Figure 6.

Figure 6:

The Motivation and Sedimentation Model of meaning making, with upward motivation relations and downward sedimentation relation (from Devylder and Zlatev In press).

The embodied level involves non-linguistic, cognitive, and experiential processes and structures such as bodily mimesis (Donald 1998), emotions (Foolen et al. 2012), categorization (Rosch 1977), and analogy-making (Gentner and Markman 1997). The latter corresponds to a form of (diagrammatic) iconicity (Devylder 2018; Itkonen 2005; Jakobson 1965; Zlatev 2016). As capacities, these processes are universal, part of our human nature, although they can be shaped into culture and context-specific manifestations through interactions with the other two levels. The sedimented level consists of social and linguistic norms (Itkonen 2008a), writing systems and symbolic notations (Goody 1977; Donald 1998), and shared cultural beliefs, histories, and ideologies (Forceville 2017). The situated level is the most dynamic, creative, and tightly dependent on the immediate context and ongoing social interaction. It is the level of emergent contextual meanings, subject to interpretation and play.^[13]

The primary motivation operation is that between the embodied and the situated level, reflecting the claim that situated meaning making presupposes pre-linguistic, pan-human properties of human embodiment and features of the universal human lifeworld.^[14] In a second step, the contextual meanings become relatively consolidated over time in a “downward” direction at the sedimented level. This, in turn, co-motivates future contextual meanings at the situated level: the secondary motivation relation. Applied to metaphor, this implies that to comprehend a verbal metaphor (expression) such as (1) or a pictorial one as that in Figure 3, processes such as perceptual categorization, analogy-making, and emotional appraisal are necessary as primary motivators. However, they are not sufficient, as metaphorical interpretation within and across semiotic systems is always co-motivated by the sedimented level of socio-cultural knowledge as well.

Even this quite schematic description of the model allows us to draw a number of implications for the various debates concerning metaphor. First, MSM implies that metaphors are fundamentally signs in a particular semiotic system, rather than mappings or any other kind of cognitive correspondences and structures. The latter, like the hypothetical LIFE IS GAMBLING GAME diagram in Table 1, can rather function as part of the motivation for the situated use of metaphors like (1), given here once more for convenience, which eventually leads to their conventionalization on the sedimented level.

Second, and in line with the semiotic tradition, MSM implies that metaphors are primarily iconic (e.g., resemblance-based) signs, but not like images or onomatopoetic expressions where the resemblance is between expression and content, but between the contents of two signs: a more directly given (“source”), and a more relevant for the context (“target”). With respect to (1), these are respectively the literal senses of terms like winner, takes-it-all, loser, small, and their metaphorical counterparts, or more generally: between the experiences of winning and losing in a game, and corresponding experiences in life. Concerning the GREEK FLAG ≡ TOILET PAPER pictorial metaphor in Figure 3, these interpretations are toilet paper (with its associations) and, with the help of metonymy (FLAG FOR NATION), Greece. In all cases, the similarity between source and target meanings co-exists with various degrees of tension, from mild ambiguity to obvious incongruity.

Third, the model implies a scale of metaphoricity: a particular metaphorical expression will be experienced as metaphorical in proportion to (a) the degree of iconicity and (b) the degree of tension between the source and target concepts. If a metaphor is only motivated by the sedimented level, that means that it has been so conventionalized that neither (a) nor (b) will be prominent. Thus, expressions like (2) may be taken, in a particular cultural context, as quite literal.

In accordance with these implications, the definition of metaphor proposed by Devylder and Zlatev in press is the following: an expression in a given semiotic system (or a combination of systems) with (a) at least two different potential interpretations (tension), (b) standing in an iconic relationship with each other, where (c) one interpretation is more relevant in the communicative context, and (d) can be understood in part by comparison with the less relevant interpretation.

Corresponding to this theoretical definition of metaphor the reader should have noticed that we have arrived at a conception that is in several ways a synthesis of the cognitive linguistic and semiotic traditions reviewed in Section 2. While we agree with semiotics that metaphors are fundamentally signs, we accept as plausible the existence of analogies on either the embodied level (“primary metaphors”) or on the sedimented level (“complex metaphors”) as motivations for the use of metaphorical signs (expressions) on the situated level. Concerning the debate on universality vs. culture-specificity, acknowledging the co-motivating powers of the embodied and sedimented levels allows for various combinations of both factors. Critiques of CMT that targeted its relatively static mappings can be addressed by pointing to the fundamentally dynamic and context-sensitive nature of the situated level. Thus, we agree with Kolter et al. (2012: 221) that metaphors “should be regarded as a process of meaning construal in which new metaphoric expressions dynamically emerge, are elaborated, and are selectively activated over the course of a conversation,” and at the same time emphasise that these would not be possible without universal processes of meaning making on the embodied level, and cultural conventions and shared knowledge on the sedimented level. Furthermore, the model highlights that metaphors can be expressed in semiotic systems other than language, such as gesture and depiction, and also through the integration of such systems in polysemiotic communication, as we illustrate in Section 4.

Finally, the model also addresses what is perhaps the most hotly contested issue in metaphor theory today: the role of consciousness. With its roots in phenomenology, MSM implies that human meaning making is always more or less accessible to conscious awareness, even when it is not in focal consciousness, but in its margins (Zlatev 2018). Even the “unconscious,” from the perspective of phenomenology, leaves its traces on conscious experience (Brooke 1986; Mouratidou 2019). In short, consciousness, as the “depth-structure of subjectivity” (Zahavi 1999: 206), is a scalar concept, varying from unconscious and pre-conscious to semi-conscious and focally conscious experience (Zahavi 1999; Sonesson 2009).

In relation to metaphor, this implies that there is no fully unconscious metaphor use, as the two properties that we propose as definitional — iconicity and tension — need to be experienced by both metaphor producers and interpreters. This, of course, does not mean that all metaphors should be produced with the explicit intention to be understood as such, e.g., deliberately, as this is an additional dimension of metaphor (e.g., Steen 2009, Steen 2017). Quite a few, especially among (verbo-) pictorial metaphors that are most often highly creative, will be of this kind, as highlighted by Forceville (2009: 23, our emphasis): “a phenomenon to be labelled a (visual) metaphor should be understandable as an intended violation of codes of representation, rather than as being due to carelessness or error” and also strongly supported by a number of ethnographic go-along interviews conducted with Athens-based street artists periodically between 2015 and 2018 by Stampoulidis.^[15] Thus, we maintain that metaphor use is fundamentally linked to human consciousness, to a much greater extent than to the “automatic mind” (Gibbs and Chen 2018).^[16]

In the following section, we show how this cognitive semiotic framework for metaphor both guides and helps interpret the findings of a recent empirical study of (verbo-) pictorial metaphors in Greek street art.