Image schemas in narrative macrostructure

CHAPTER EIGHT IMAGE SCHEMAS IN NARRATIVE MACROSTRUCTURE: COMBINING COGNITIVE LINGUISTIC WITH PSYCHOLINGUISTIC APPROACHES MICHAEL KIMMEL Abstract In how readers represent narrative plot-macrostructure extended imaginative gestalts may play a key role (parallel to or instead of more abstract propositions). Image schemas, a notion employed very successfully by cognitive linguists (Gibbs & Colston 1995, Hampe 2005), are an excellent tool for modeling macrostructure in narratives. Applying this hypothesis requires projecting claims about phrases to more extended events and the skeletal representations that readers create of them. As this is methodologically all but trivial, how can we explore if story ontology really involves complex gestalts with image-schematic and sensorimotor properties? Clearly, combining text-linguistic with psycholinguistic methods is highly advisable here. The first major section of this paper summarizes a textual approach to Joseph Conrad’s novel Heart of Darkness, from which I generated hypotheses about the relation between FORCE, PATH, CONTAINER, and other image schemas on the one and specific “tracks” of event ontology on the other hand. The second section reports on a priming experiment with 40 short stories, which focuses on the “track” of story causality. It was hypothesized that abstract representations of story causality (successful vs. unsuccessful goal completion) involve FORCE-like aspects. In the experiment, readers had to perform dynamic force gestures with their hands that could potentially speed up the comprehension of plots with similar flow contours. 2 Image Schemas in Narrative Macrostructure Introduction I would like to begin this essay on global story representations with a couple of local textual passages, which I culled from Joseph Conrad’s famous novella “Heart of Darkness”. „I went a little farther (...), then still a little farther, till I had gone so far that I don‘t know how I‘ll ever get back“ (p. 90) „fantastic invasion” (p. 58), “fantastic intrusion“ (p. 95); „tear treasures out of the bowel of the land (...) burglars breaking into a safe“ (p. 55) „deeper and deeper into the Heart of Darkness“ (p. 62); „travelling back to the earliest beginnings of the world“ (p. 59) To see how we can make the transition from local text to global story representation, let us ask what these quotes distributed across the text have in common. First, all passages seem to hold a key status in cueing the reader to understand the story’s manifest event structure and - when one goes into interpretation - deeper layers of meaning as well. Second, all use spatial and sensory imagery for abstract facets of the story event (even in the case of the first quote which could be read as spatial, but also refers to a psychic “movement”). Third, it is striking that the imagery of the various text passages seems to amalgamate easily at a conceptual level. The expressions evoke (a) path, which they specify (b) as being covered by applying force, and (c) as leading into some kind of inner-realm or depth. Thus, all the expressions (together with three or four dozen others that I don’t mention here for the sake of brevity) seem to hang together in the sense of specifying complementary facets of story progression. They all help the reader create a mental model for understanding the deeper psychological and metaphysical nature of Marlow’s disconcerting riverboat journey into the Congo in search of the enigmatic Mr. Kurtz. This material is drawn from a case study that applies cognitive linguistic methods to literary cognition (Kimmel 2005). In it, I have examined how metaphors and other imagery-imbued words may cue the reader’s understanding of the global action and event structure of the novella. I have also looked at how distributed cues of an imagistic sort may reflect subtler literary and aesthetic effects which piggyback on the understanding of the basic event structure. These include symbolic nodes (e.g., the rich symbolism of what “darkness” signifies), megametaphors (e.g., the journey as a parable), and the reader’s sensory-motor involvement in the reading process. In the present chapter, I will use this material as a starting point to illustrate a more general cognitive theory of story macrostructure. In essence, my claim is that a story’s macrostructure itself shares with the Michael Kimmel 3 above metaphors something gestalt-like and imagistic. It is not only in Heart of Darkness that the mental model of the overall story ontology may involve the same gestalt aspects (PATH, INSIDE-OUTSIDE, CENTERPERIPHERY, DEPTH, FORCE-PENETRATION, and the like) found the local text-passages. I claim that gestalt scaffolds, onto which metaphoric and other language may open a window, are a general feature of comprehending and recalling stories. The first purpose of this essay is to explain this hypothesis from a text-linguistic and literary viewpoint and then to go into a discussion of a psycholinguistic experiment with simpler stories that validates the claim. 1. Narrative macrostructure Before I outline my own theory, I will define the key notion. The term narrative or story macrostructure – related to the terms “theme”, “plot” and “storyline” – refers to a compressed representation that arises when a reader selects, combines and makes coherent globally relevant aspects from the text microstructure, usually augmented by schematic knowledge (e.g. of genre schemas) (van Dijk 1980, Vipond 1980). Macrostructures thus create a skeleton-like summary representation of an event’s known or 1 expected structure. Generating macrostructures has been described as an inferential process that reduces information (Kintsch 1993). Macrostructures “selectively encode input that addresses the emerging theme […] an information-rich, compact representation that permits the reader to construct characteristics of the text from multiple perspectives” and produce “adages that succinctly capture conflicts, planning failures, solutions, and resolutions” (Graesser, Pomeroy, & Craig 2002: 30, 26). Macrostructures fulfil various cognitive functions in story comprehension: First, they let readers represent the theme of a story and its storyline as a more or less coherent meaning structure and allow creating explicit verbal summarisations of a story’s gist. Creating this kind of compressed representation is crucial because the recall of long texts never happens as a one-to-one reproduction of the text base, but requires a summarisation process through which microstructures are deleted or 1 A macrostructure may be conceived of as an extended situation model or as a structure developed from a collection of situation models. Situation models are the term chosen by discourse psychologists for mental microworlds whereby readers represent the actions, events, etc. of a text. They result from the integration of episodic text memory with prior domain knowledge (cf. Ferstl & Kintsch 1999: 247). 4 Image Schemas in Narrative Macrostructure reconstructed. The main purpose of successful macrostructures is that they condense the possible inferences readers may make about a story. Macropropositions that summarise a story well seem to have a special status in event memory (Guindon & Kintsch 1984) and allow of a range of inferences close to those of the complete story (Graesser, Bowers, Olde, White & Person 1999). Second, in on-line story comprehension the tentative model of a macrostructure at a given moment will guide the reader in selecting, making salient and rendering coherent further input from the text. Thus any macrostructure is an emergent and dynamically evolving model of story ontology, which results because the reader makes an active effort at achieving coherence. Third, cultural knowledge flows into constructing macrostructures, especially of story genres, of prototypical or “good” stories and of everyday scripts. Readers recognize familiar story models like the exposition-complication-resolution structure (Bartlett 1932), but may also elaborate them further or create a salient contrast with them. Familiar schemas like story genres or those of more particular story prototypes facilitate macro-comprehension (Kintsch & Yarborough 1982). Thus, readers are guided by both world knowledge and implicit contextual expectations in their strategy of fitting incoming text into an expected macrostructure (cf. Petterson 2002). 2. Image schemas in narrative macrostructure After this prefatory sketch, I may now lay out what is a relatively new view of macrostructure. My hypothesis is that readers represent important parts of plot-structure as an extended, dynamic, imaginative and gestaltlike contour. The theory of image schemas (Gibbs & Colston 1995, Kimmel 2002, Peña 2003, Oakley 2004, Hampe 2005) provides the conceptual tools to make sense of various aspects of skeletal event representations with such gestalt-like properties. A benefit of the framework is that both text-linguistic and a psycholinguistic research on image-schematic aspects of events has been done in recent times, although it has not extensively been applied to narrative macrostructure. An exception is my own text-linguistic research (Kimmel 2005) and the experiment I report on below. My claim that image schemas underlie event/story comprehension counters a long tradition of modelling macrostructures in a non-analogue and propositional format (e.g., van Dijk & Kintsch 1983). This tradition defines macrostructure as a complex hierarchy of propositions, and while Michael Kimmel 5 it explains well how information is compressed (namely by selecting, deleting and recombining propositions), it has considerable difficulty in explaining both the grounding of story comprehension in percept-near cognition and embodied processes in reading. My alternative to this is the suggestion that macro-gestalts account for global story structure. These macro-gestalts can be described as image schemas. 2.1 Image schemas defined Image schemas, according to Johnson (1987), are experientially recurrent non-propositional representations which are distinct from rich (= percept-near) images. Rather, they are mere topological scaffolds that inhere as dynamic structuring principles in percepts, conceptual activity, and action patterns. Image schemas are cross-modal and connect kinesthetic, tactile, visual and auditory imagery and interface between the embodied and the conceptual realm. Developmentally, they constitute the embodied protosynthesis of abstract concepts in spatial and kinesthetic experience (Mandler 1992). Typical image-schematic gestalts, usually written in small caps, include CONTAINER, BALANCE, CYCLE, PATH, CONTACT, UP-DOWN, LEFTRIGHT, CENTER-PERIPHERY, PART-WHOLE, FORCE, SCALE, NEAR-FAR, STRAIGHT, or MULTIPLEX-MASS. What is important here is that many simple image schemas, when dynamically construed, already represent a minimal event. When a path is followed, something is raised up, a boundary transgressed, a circle traced, an object split, or two objects merged, when the perceptual focus wanders from left to right or from part to whole or “zooms in” from mass to multiplex in each case this constitutes a mini-event with a recognisable gestalt structure. Each gestalt already constitutes a simple conceptual scaffold that otherwise diverse kinds of events share. More complex events involve image schemas in sequence, with a change of manner or direction. For example, a mental representation of “anger” involves a mini-scenario in which a container (the body) comes under pressure from within, which goes to a maximum, upon which a sudden release deflates the container again (Lakoff & Kövecses 1987). This is a sequence of EXPANSION and CONTRACTION of a CONTAINER driven by increasing FORCE from within and thus a scenario constituted by compound and dynamically evolving image schemas. 6 Image Schemas in Narrative Macrostructure 2.2 Image schematic event structure in cognitive linguistics and cognitive narratology The reader may by now have guessed that the quotations from Heart of Darkness on the opening page in fact were chosen for the fact that all of them exemplify image-schematic expressions which hint at some aspect of how the event of traveling up-river in search of Kurtz unfolds. That linguistic expressions can be used to infer image-schematic gestalts associated with them is one of the fundamental assumptions of cognitive linguistics. Evidence for image schemas in event cognition comes from linguistic research about word and phrases, some of which I summarise here, as well as a growing number of corresponding psycholinguistic experiments (summarised in Gibbs 2005). First, there is evidence on event shape. The manner of an event is encoded in lexical or grammatical aspect which can be perfective, imperfective, habitual, continuous, progressive or stative. Spivey et al. (2005: 269) suggest that perfectivity has to do with image-schematic boundedness. Parrill’s (2000: 20) study suggest that features of verb aspect such as dynamicity (change over time), closure (boundedness in time, or telicity), iterativity (repetition) and durativity (extension in time) “map nicely onto actual observable features of physical motion” in the accompanying gestures used by speakers. Gestural and grammatical aspect apparently share a common origin, e.g. perfectives are frequently accompanied by single emphatic chops and ballistic slaps to mark the clear conceptual boundary that reflects completed action. Studies on cross-modal attunement between mother and child also offer insights about event shapes (Stern 1985). For example, a soothing prosody of “THEre-THEre” shares a cross-modal contour with the way the pressure of the accompanying caress is softened 2 at the end of each movement. A second major aspect that is imageschematic in events is what drives an agent. For example, Sweetser (1990) demonstrated that force-dynamic image schemas underlie modal verbs. In 2 Apparently, image schemas also determine the meaning of some verbs such as pushing, pulling, resisting, yielding, releasing, dipping, rising, climbing, pouring, falling (Talmy 2000). Narayanan (1997) claims that verb semantics are defined by “physical motor control primitives such as goal, periodicity, iteration, final state, duration and parameters such as force and effort” (quoted from Zacks & Tversky 2001: 11). The different kinds of processes expressed by verbs can be categorized according to whether they are (1) internally homogeneous or heterogeneous, (2) temporally bounded or unbounded, and (3) involve energy consumption or not. This corresponds roughly to CONTAINER, MULTIPLEX-MASS (or BOUNDEDNESS and CONTAINER) and FORCE schemas. Michael Kimmel 7 “root modals” a real person is allowed to or required to do something, such that these modals can be based in prototypical FORCE configurations experienced in space. Sweetser claims that “epistemic modals” that emerge from them by metaphorical extension involve similar force representations. For example, while “may” in “You may be right” denotes the ABSENCE OF AN EXTERNAL BARRIER TO FORCE, “must” in “He must be the Scarlet Pimpernel” denotes an IRRESISTIBLE FORCE MOTION, and “cannot” in “She can’t have gone over to the enemy” denotes ENABLEMENT OF A MOTION IS ABSENT. Some recent approaches in cognitive narratology transpose imageschematic structure to a more extended time-scale and claim that narrative has a temporal “event shape” and other aspects similar to the ones found in sentences. Johnson (1993: 69) claims that narrative tension involves image schemas of balance that gets upset and may be restored. Talmy (2000: 439) suggests that force dynamics in narrative plot “characterize such relationships as two entities opposing each other, a shift in the balance of strength between the entities, and an eventual overcoming of one entity by the other.” Most systematically, Turner’s (1996) study of literary metaphor and parable claims narrative and poetry is structured by “small spatial stories” that are “routinely held together by one or more dynamic image schemas” (p.19). He suggests that we think of “events in time, which have no spatial shape, as having features of spatial shapes – continuity, extension, discreteness, completion, open-endedness, circularity, partwhole relations, and so on” (p. 18). More specifically, Turner claims that we image-schematically understand an event’s internal structure such that it “can be punctual or drawn out; single or repeating; closed or open; preserving, creating, or destroying entities; cyclic or not cyclic, and so on.” (p. 28). In addition to this event shape, causal structures are also understood by projecting onto them image schemas of force dynamics and of movement along a path (like in the sentence, ‘Fear drove him to a situation he otherwise would have avoided.’). In some instances, causal relations in an event may also be thought of as links, paths, or emergence (p. 18). Furthermore, drawing on Sweetser’s work, Turner describes narrative event modality as using force image-schematic. This includes an agent’s ability to perform an action, the obligation or necessity to perform it, or the possibility of some condition allowing the actor to perform it (p. 29). 8 Image Schemas in Narrative Macrostructure 2.3 Towards a typology of narrative image schemas The recent work on image schemas has prompted me to further systematise the major types of image schemas that may structure narratives at various levels. Authors have overlooked so far that imageschematic models may structure events and literary cognition at various levels, which we need to hold apart (cf. Kimmel, n.d.): (1) Local image schemas recur in local text, most often in metaphors, and thereby become “thematic”. Examples would be locally recurrent BALANCE relations or CONTAINERS (e.g. Stockwell 2002, Freeman 1993, 1995) that structure several metaphors that play a major role for the plot. Although these image schemas are repetitive and may of course create major memory pegs for the plot summary they do not genuinely inform the understanding of episodic progression itself. The following three types are more manifestly macro-structural. (2) A second category represents temporal and hierarchical story structure as PARTS AND WHOLES, NESTED subgoal structure, episodic breakpoint INTERVALS, etc. This category is mostly concerned with organising meso-structure. (3) A similar kind of image schemas is used by readers to construe what happens in action-oriented terms (the plot) in terms of causality and protagonist interaction (FORCES). Again, this category is mostly organises the episode level of a story. (4) A high-level type of image schemas holistically captures the changes over the PATH time-line of an event. These arise as emergent structure when then interrelation of local information is globally conceived, much like a melody in music. Such holistic image schemas can be felt by readers in their mind-bodies when they represent the story’s affect contour, for example as an arc of tension that ends in a resolution or a restoration of balance. In developing psycholinguistic experiments we will have to focus on types 2-4, which capture the progression of an episode or entire story, hence a real macro-structural effect (although 4 also seems difficult to deal with because of its highly encompassing nature). 3. A text-linguistic approach to macrostructure based on image schemas We need to account for how a given story lets several kinds of imageschematic meaning unfold in parallel, because only various “story track” representations together create a full narrative. For example, Zwaan, Magliano and Graesser (1995) assume that readers must encode, monitor and update spatial, causal, intentional, temporal, protagonist, and Michael Kimmel 9 emotional dimensions during reading, notably – as various studies show – causal, intentional and protagonist information. Although the tracks in my model do not wholly coincide with this framework, I agree that a full macrostructural representation will compress information with regard to several tracks. Predicting which kind of story substructure creates which kind of image-schematic imprint is not only needed for fleshing out the image schema model of reading. It is also a crucial condition for designing experiments, because each experiment can only target one track at a time and it needs to be clear which text cues evoke it. I will now enter into the details my study of metaphors in “Heart of Darkness” with a focus on the data pointing to event ontology (Kimmel 2005). First, some words about the novel’s plot. Marlow, a seaman, recounts his riverboat journey into the depths of the Congo to find the enigmatic Mr. Kurtz who has established an irrational reign of charisma and terror at his trading outpost, but now is at the verge of death and madness. The gradual intrusion into the eeriness of the jungle transforms Marlow‘s senses and the confrontation with Kurtz his soul. He experiences a kind of initiation into “dark” knowledge. Back in “civilisation” Marlow visits Kurtz‘s fiancée, but keeps the truth about his fall from grace and his last words (“The horror! The horror!”) from her. I will mainly look at the event-ontology related metaphors that characterise the main stage of the journey. These can be grouped into different kinds of ontological aspects. For each type of “track” I will now graphically depict the schematic constituents together with a selection of textual metaphors that cue the representation. 3.1 Ontological story spaces One kind of image-schematic structuring concerns the spatial layout of a story scene as CONTAINER, CENTER-PERIPHERY, LANDMARK, CONTACT, APART, NEAR or DISTANT etc. Since models of single spaces like rooms, etc. are most characteristic of the single episode level, they tend not to impact the macrostructure too much. The narratively most interesting kind of mental representation concerns spaces in a more abstract or even metaphorical sense. Spaces may conceptualise not only spatial locations, but also psychological and existential states represented as bounded or non-bounded containers (cf. Lakoff & Johnson 1999) superimposed on the time-path a protagonist moves through. In Heart of Darkness Marlow, the protagonist, moves from the European space of restrained passions, enlightenment and “culture” to another deeper space, the jungle wilderness, that is associated with 10 Image Schemas in Narrative Macrostructure attributes of passion, irrationality, the dark, and the feminine. Here we have a logical pattern of the utter apartness of spheres, of ontological incommensurability (indicated by bundles of opposing attributes). This is bolstered by a moral, metaphysical and sensory opposition (light / dark, ...) between the two spheres. out-space “civilization“ in-space “wilderness“ “ Figure 1 Evaluative and metaphysical opposition as APART, DIFFERENCE „blank space of delightful mystery“ [the Congo] (p. 22); „that great and saving illusion“ [the fiancée’s naive world] (p. 121); darkness and light metaphors (Kurtz as torch bearer of enlightenment, etc.) Spheres are related as OUTSIDE-INSIDE or SHALLOW-DEEP Concentric trip: Outer Station – Central Station – Inner Station „Mr. Kurtz was in there... a little ivory coming out from there” (p. 49); „deeper and deeper into the Heart of Darkness“ (p. 62); „traveling back to the earliest beginnings of the world“ (p. 59) 3.2 Event modality and temporal texture The temporal nature and manner of an event is connected to what was earlier said about “aspectual classes” in grammar. Although stories as wholes are, of course, inherently extended and never punctual acts which are during reading construed as imperfective (and after reading as perfective), the manner in which key actions within the whole happen involves “event shapes” of all sorts that contribute incrementally to the whole. In Heart of Darkness Marlow’s journey unfolds as a movement along the river from one sphere to the other, and concomitantly from one psychic state to the other. The temporal modality of the movement is gradual and Michael Kimmel 11 creates a slow continuous transition between states that is accompanied by a gradual accumulation of tension. Note that the slowness in time has to do with other aspects like counterforces being present. . Figure 2 Gradual transition (FORCE-PATH, SCALE) „The grimy beetle crawled on“; „reality...it fades“ (p. 60); „I went a little farther (...), then still a little farther, till I had gone so far that I don‘t know how I‘ll ever get back“ (p. 90) Intrusion and transgression (FORCE PENETRATION) „fantastic invasion” (p. 58), “fantastic intrusion“ (p. 95); „tear treasures out of the bowel of the land (...) burglars breaking into a safe“ (p. 55) A further aspect of temporal texture is episodic structure. The breakpoints between episodic meaning units (Zacks & Tvesky 2001) may correspond to container or path-interval image schemas, such that the episode is a part in a story whole. In Heart of Darkness this is forcefully expressed in the language of doors, edges and threshold (see below).3 3.3 Plot-driving agency and agent intentionality Protagonist agency pertains to where the impetus for an action comes from. Talmy (2000) specifies that agents have an inherent force tendency. Notably, an agent is conceived of as endowed with a force emanating 3 Event texture over time may produce emergent perceptions of story “speed”, emotional tension, density of action or the like. A contour over time can be GRADUAL vs. ABRUPT, INCREMENT vs. DECREASE, ACCELERATION vs. DECELERATION, MAINTENANCE. Stories can be STRUCTURED or ERRATIC, as well as STATIC, CYCLIC or CHANGING. However, this kind of effect goes beyond what any metaphor can express. 12 Image Schemas in Narrative Macrostructure either from her intrinsic motivation (intrisic force agency) or from an outer or extrinsic causality (external force agency). Intrinsic force agency may be further conceived of as FORCE ENABLEMENT when it is not yet enacted and as FORCE BLOCKAGE REMOVAL when something needs to be done first. Extrinsic force agency can be conceived of as FORCE PULL, FORCE ATTRACTION, FORCE IMPACT, or OBJECT DESTRUCTION BY FORCE that drives the agent entity or enables her actions by removing a blockage, etc. In Heart of Darkness, the protagonist Marlow (and before him Kurtz, whom he emulates) moves into Africa driven not only by his task for the trading company, but, as many metaphors indicate, by a kind of desire and curiosity, a force of internal agency. At the same time, the wilderness is a quasi-agent that actively exerts ATTRACTION, a force of external agency. force attractor driving force Figure 3 Wilderness [a quasi agent] as seductress and attraction (FORCE ATTRACTOR) „smiling... inviting, mute with a air of whispering, Come and find out.“ (p. 29); „beguiled his unlawful soul beyond the bounds of permitted aspirations“ (p. 107) Desire as impelling force (FORCE DRIVE) „memory of gratified and monstrous passions … had driven him out“ (p. 107) 3.4 Causality An aspect related to the previous one is how events are causally connected within and between episodes. This most often concerns how the main goal of a protagonist is attained over various stages and subgoals (cf. Mandler 1984). In terms of image schemas, goals are PATH ENDPOINTS (Lakoff & Johnson 1999, Johnson 1993: 68) and causality consists of maintained, frozen or broken FORCE CHAINS toward this endpoint, which Michael Kimmel 13 pass on a force impetus. Langacker (1987) calls this the “billiard ball model” of causality. The force-chains are superimposed on the PATH-like timeline. In causal models, hierarchical substructure may issue when a protagonist has to achieve a subgoal. This may be understood as a NESTING-relation, a particular kind of PART-WHOLE schema. Thus, nested force-chains may occur, before the main force-chain can be resumed. In Heart of Darkness explicit metaphors for macro-causality (in the sense of how one event leads to the next) are comparatively absent. However, the continuous FORCE ATTRACTION and FORCE-PULL of agent intentionality that moves Marlow towards the endpoint are established from the beginning on (Kurtz, the Inner Station) and account for the overall action chain without further cues. 3.5 Obstacles and overcoming them A special facet related to agency/intentionality and causality has to do with the presence (or absence) of obstacles in reaching a protagonist goal. It has to do with FORCE BLOCKAGE and FORCE BLOCKAGE REMOVAL. (In a sense forces that block a story’s progress can be more or less treated on a par with antagonists, which I discuss separately here, however.) Despite the forces that drive the protagonist onward, there are both moral and physical barriers on the journey, the moral ones being related to Marlow‘s inner turmoil. Yet, the attraction and “gravity” help overcome the barrier (curiosity, captivation with Kurtz), which results in entering the other sphere and a kind of initiation into the ‘Other’ in a quasi-ritual process. 14 Image Schemas in Narrative Macrostructure barr ier Figure 4 Reaching boundary / threshold (FORCE, PATH, CONTAINER) „guarding the door of Darkness“ (p. 26); „skirts of the unknown“ (p. 61); „toiled along slowly on the edge“ (p. 62), „peeped over the edge“ (p. 119) Barriers (FORCE BLOCKAGE) Physical: tropical climate, attacks, death of companions; Moral: subtly insinuated when Marlow witnesses cruelties; Psychic: fear and madness Transgression (FORCE BLOCKAGE REMOVAL) „beguiled his soul beyond the bounds of permitted aspirations“, „driven him out to the edge of the forest“ (p. 107); „Transgression, punishment – bang!“ (p. 48) 3.6 Protagonist interaction At an interactional level, what narratologists after Greimas (1966) analyse under the heading of “actant” roles is represented as force configurations between agents. Thus, a protagonist’s role at a given time may involve forcing, letting, preventing, helping, hindering, another agent or object as well as acting in vain, each with a specific force-dynamic schema configuration (i.e. FORCE ON OTHER, FORCE RESTRAINT REMOVAL UNTO OTHER, COUNTERFORCE AGAINST OTHER, CONVERGENT FORCE WITH OTHER, FORCE BLOCKAGE OF OTHER’S MOVEMENT and FORCE THAT FAILS TO MOVE AN OBJECT). Patterns of vying forces between protagonists may also change over time, e.g. a shift of the balance of strength, of overcoming the other, or a subsiding antagonism at the end of a story when a complication is resolved and a denouement ensues. In Heart of Darkness a recurrent subtext is the interplay of the Europeans with an antagonist, the wilderness, which has all the attributes of an agent and thus internal force agency. Conrad creates an explicit uncertainty as to whether “white man dominates Africa or the converse”. The metaphors used point to an interplay of forces (which is, probably not Michael Kimmel 15 by coincidence, also the canonical conceptual model for the relation between emotions and rationality, cf. Kövecses 2000). Figure 5 Wilderness as an agent and enemy (FORCE ANTAGONISM, VYING FORCES) Metaphors of wilderness dominator, avenger, animal, etc. (see above); „taking possession of an accursed inheritance“ (p. 62); „Could we handle that dumb thing or would it handle us?“ (p. 49) Furthermore, there is a process in which Marlow gets taken in by wilderness and is thereby transformed in his soul. This is presumably effected because the wilderness exacts a force on him. Figure 6 Captivation / being taken in (FORCE ENGULFMENT, CONTAINMENT) „feel the savagery... had closed around him“ (p. 19); „wilderness took him into his bosom again“ (p. 45) „being captured by the incredible which is the very essence of their dream (p. 50)“; „shadowy embrace“ (p. 100), „powers of darkness claimed him for their own“ 3.7 The compound plot model and the literary effects that it produces Strikingly, these various metaphor-cued aspects of plot-imagery neatly fit into a single compound image schema. Thus, the conceptual spaces, vectors, and entities arising from several kinds of metaphor can be superimposed imagistically in a single mental substrate where they specify each other. The conceptual base which is responsible for letting aspects of 16 Image Schemas in Narrative Macrostructure imagery become co-specifying may thus be called a “mental sketchpad” (cf. Baddeley 1986). force attractor out-space „civilization“ barr ier in-space „wilderness“ journey driving force Figure 7 Something interesting may follow from this model of literary processing: When local features become coherent in a global representation by virtue of their image-schematic affinities, further inferences are invited by their combined constraints. In the case of Heart of Darkness the compound, integrated and relational nature of this imageschematic skeleton has important inferential and aesthetic properties for literary cognition. For example, the pulling and pushing forces, together with the implied barrier, create a very specific effect of literary tension. Likewise, the mode of gradual transition together with an intensification of attraction make the in-space of the Inner Station as well as Kurtz more and more electrifying. Summing up, I have concluded: “readers activate imagery lattices to monitor potential information overlaps between text segments, to link metaphors and cultural models into a higher-level structure, and to produce novel inference. In other words, readers actively search for and mentally simulate matches between imagistic story features they have read” (Kimmel 2005: 233) What does this metaphor-based case study then reveal about eventrelated cognition in literature? I do not mean to suggest that every text contains so wonderfully explicit clues of the overall image-schematic event ontology. Some texts do, and others will rely more on the implicit knowledge the reader has about story structure and, in particular, genre Michael Kimmel 17 knowledge. However, if my image-schematic hypothesis is correct a further-reaching claim can be made: Even where the macrostructure cannot developed in every aspect from metaphor analysis, the underlying mental model that readers eventually construct of the story ontology will share its cognitive features with more explicit cases. Therefore, I will next take a closer look at stories which lack explicit metaphoric cueing by discussing a psycholinguistic experiment on emergent image schemas in plot-comprehension. 4. An experimental approach to macrostructure To validate this model of gestalt macrostructures experimentally I chose as a story “track” the force dynamic aspect of causality and goalattainment. As stimuli I created stories that systematically vary this aspect of plot. The experiment took its inspiration from several related linguistic studies, which I lack the space to summarize here (but see Gibbs 2005: 174-207), but which have produced mounting evidence for the claim that image schemas structure abstract events at the phrase-level. 4.1 Motor priming experiment The experiment, conducted jointly with Barbara Kaup of the Technical University of Berlin, aimed to find out whether the suggestive phrase-level findings may be extended to the higher level of story macrostructure. A motor priming design was used in the interest of studying the on-line activation of image schemas in short story reading. Because image schemas are believed to be cross-modal,4 my guiding assumption was this: The cognitive planning effort to carry out a force-dynamic gesture (i.e. activating a motor representation) can interact with a force-dynamic structure that is hypothesized to be part of the reader representation of story goal-attainment. On the basis of this carry-over, a priming design was created such that matches between plot and manifest gestural motorprimes were predicted to let the reader create a macro-representation more quickly than mismatches. The build-up of this representation we were able to deduce from reading times for key sentences. The chosen “story track” agent-driven causality in goal-attainment could manifest itself in three 4 This adds to the assumption that stories have a gestalt-like core. The experimental design implies that this gestalt is also cross-modal and thus not divorced from bodily feeling, but involves substantial sensorimotor and proprioceptive dimensions. This expectation is well rooted in cognitive linguistic research. 18 Image Schemas in Narrative Macrostructure kinds of force-dynamic schemas, which in turn reflect three kinds of plot dynamics. Story stimuli. The stories were “textoids” constructed to ensure that every plot featured a main protagonist with a clear goal, a low number of other agents, no digressions or embedded sub-episodes across test and control conditions. The stories were short, fairly attention-grabbing, and had an invariant kind of plot progression underneath divergent surface structures: (1) A protagonist is introduced in the first sentence with a clear main goal, the achievement or failure of which is at the same time the main topic of the story. Depending on the experimental condition the goal is either reached without trouble, after some difficulty, or not at all. (2) While other agents occur especially in the roles of helper and antagonist, emotion cues and perspectivization devices ensure that the reader takes the perspective of the main protagonist. The rationale for this is that the macrostructure representation may be, to some degree, perspectival in the sense that identification with another person may lead to another kind of causal model, etc. (3) Care was taken that no words evoking strong force dynamic schemas occurred in the text to avoid a priming effect rooted in the manifest microstructure rather than the inferred macrostructure. (4) Although actual spatial movement by protagonists is impossible to exclude, care was taken that the protagonist goal is no genuinely spatial goal and that achieving it did, for the most part, require abstract movement (like a thought-path, etc). Gestural primes. For each of the 40 stories, one of four simple gestures had to be performed. Beginning from a position of both hands held at elbow-height and at shoulder-breadth one of the following was instructed before each story: (1) the left fist moves forcefully to the clasping right hand, which stops its movement through counterforce; (2) the left fist moves to the right unstopped by anything as far as the impetus goes; (3) the fist is first briefly stopped by the right hand, but then it is pushed away and the movement continued; (4) an outward opening motion of both hands with both palms up is done (filler gesture). Each of these four gestures matched an icon: Figure 8 Michael Kimmel 19 Procedure. The 46 participants were asked to read forty short stories of 10-12 sentences each in a self-paced fashion from a laptop screen, moving forward between screens by pressing the space bar. After familiarising the subjects with the reading and gesturing tasks they read one story after the other. Before a story began one of the four visual icons was displayed so that the subjects could memorise the associated gesture and later carry it out upon hearing a beep during reading. In the final debriefing the participants were asked if they had a clue about the research question and whether they had consciously reflected on a connection between gestures and story plots. (The former was never the case, the latter occasionally.) Each experiment lasted between 20 and 40 minutes. Test design. Cognitively, the motor memory of the gesture had to be held active in parallel to on-line text comprehension. Our aim was to ascertain whether the (mis-)match between the planned gesture and the story would influence the reading speed of three target sentences in which the story complication occurred. These sentences were located towards the end of each story, but never occurred in last or penultimate position. The reading speed was deduced from the self-paced space-bar presses. A tri-factorial 2(group) x 2(condition) x 2(match) design was used. Every participant read two different groups of stories (see below for details) and did so including all the conditions. The groups comprised different stories. Twelve stories were included in group 1, another twelve in group 2, and sixteen used as fillers. In an overview, the three factors in which manipulations occurred were: o o o The story type (12 stories in Groups 1 and Group 2 each, and 16 additional filler stories): The central part of each story sequence (2 plot versions in each story): The four gestures could either match the story plot or not. The following table specifies whether story type and gesture match and provides the number stories used in each test condition (in brackets): 20 Image Schemas in Narrative Macrostructure Table 1 STORY TYPE / PLOT VERSION TYPE 1 TYPE 2 FILLER Goal attainment / no problem Goal attainment after problem is mastered Goal attainment fails / problem persists Goal attainment after problem is mastered (gesture early in story) (gesture late in story) TOTAL # OF STORIES per gesture type GESTURE TYPE Blockage Source-pathremoval goal (unblocked) Match (3) Mismatch (3) Folding hands [filler gesture] Blockage remains Mismatch (3) Match (3) Mismatch (3) Match (3) Match (3) Mismatch (3) (3) (8) (3) (2) 9 12 10 9 The twelve stories in Group 1 involved plots in which the protagonist either reaches the goal without problems or after overcoming an abstract (non-spatial) obstacle. We predicted that these two plot alternatives would match with an unblocked punch and with a blocked punch gesture, respectively. The twelve stories in Group 2 involved plots in which, again, an obstacle had to be overcome. This was, however, now contrasted with a condition in which the aim is not reached at all and the protagonist “gets stuck”. We predicted that these alternative courses of plot development would match with a briefly stopped, but unblocked punch and with a blocked punch gesture, respectively. The two groups thus, all in all, involve three kinds of gestural image schemas that are hypothesised to match with three kind of causal plots: (1) enacted FORCE on a path that reaches its PATH END-POINT OR GOAL (immediate goal-attainment in plot), (2) enacted FORCE ON A PATH that reaches its PATH END-POINT, but only after a stage of FORCE BLOCKAGE that has to be overcome actively (goal-attainment after problem solving in plot); (3) enacted FORCE on a path that is BLOCKED and that does not reach its PATH END-POINT (no goal attainment in plot). Michael Kimmel 21 4.3 Results The dependent variables were the three sentence-reading times, the two sentences that varied with plot version and the reading times for the third target sentence that was constructed in such a way as to be identical in both story plot versions.5 Reading time analyses were conducted separately for these three dependent variables. This was done only when the participant had actually performed the instructed movement. However, as participants had quite faithfully reproduced the gestures the discarded cases were minimal. Moreover, reading times that were shorter than 500 ms or longer than 10,000 ms were omitted. For determining additional outliers, differences among the story items were considered by converting the reading times of each participant to z-scores and discarding those with a z-score deviating more than 2 standard deviations from the mean z-score of the respective item in the respective condition (3-6% of measures). The remaining data were submitted to 2(group) x 2(condition) x 2(match) ANOVA. This was done (1) with repeated measurement on all three factors in the analysis by participants and with (2) repeated measurement on the last two factors in the analysis by items. The overall ANOVA in the by-participants analysis produced a significant main effect of the group, with shorter reading times found in Group 1 (GOAL ATTAINMENT vs. BLOCKAGE REMOVAL) than in Group 2 (BLOCKAGE REMOVAL vs. PERSISTING BLOCKAGE), (F1(1,44)=11, p<.05; F2(1,22)=1.5, p=.24 The fact of speedier reading in Group 1 than Group 2 does not say anything about the hypothesis, but is noteworthy anyway. There also was a marginally significant interaction of group and match (F1(1,44)=3.5; p=.07, F2<1). To gain more information with respect to potential interactions, we analysed the reading times separately for the two groups. For Group 1 (GOAL ATTAINMENT vs. BLOCKAGE REMOVAL) no significant effects were found (all F<1.1, p>.30). However, for Group 2 (BLOCKAGE REMOVAL vs. PERSISTING BLOCKAGE), there was a significant effect in the by-participant analysis, meaning that shorter reading times were found in the match than in the mismatch condition (F1(1,44)=6.5, p<.05; F2(1,11)=3.5, p=.09). This effect directly supports the hypothesis. The relative sizes of the effect 5 Because of technical problems with the response time measurement, we only were able to get data from 24 participants for the first and the second variation sentence, slightly limiting the range of what can be said about them. For the target sentences coming last we got data for a total of 46 participants for 19 of the 24 story items of interest, but only got data for 24 participants for the remaining 5 story items. 22 Image Schemas in Narrative Macrostructure and direction can be illustrated by comparing match and mismatch conditions in each of the following four pairs of columns: 4.400,00 4.200,00 Mittelwert 4.000,00 3.800,00 3.600,00 3.400,00 3.200,00 3.000,00 avs_1. 1,00 avs_1. 2,00 avs_1. 3,00 avs_1. 4,00 avs_1. 5,00 avs_1. 6,00 avs_1. 7,00 avs_1. 8,00 Table 2: The four left columns in the table represent the results for Group 1, whereas columns 5-8 represent Group 2. Here is a summary of the specific conditions involved regarding both the type of force-dynamic schema tested and the condition of match or mismatch: Michael Kimmel 23 GROUP FORCE SCHEMA CONDITION Group 1 Group 1 Group 1 Group 1 BLOCKAGE REMOVAL BLOCKAGE REMOVAL GOAL-ATTAINMENT GOAL-ATTAINMENT Match Mismatch Match Mismatch Group 2 Group 2 Group 2 Group 2 BLOCKAGE REMOVAL BLOCKAGE REMOVAL BLOCKAGE PERSISTS BLOCKAGE PERSISTS Match Mismatch Match Mismatch 4.4 Interpretation, uncertainties and future directions The fact that at least for one experimental group (Group 2) a pattern of equivalence between story-plot structure and force-dynamic gestures emerged is suggestive for my hypothesis of image schema structure in plot causality. Two background conditions have to hold for this interpretation: (1) No propositional connections can be derived between gestures and stories. I believe that no such effect is possible. This would require explaining in what way sketchy visual images and subsequent motor planning can call up a propositional representation about a story goal and how transmodal correspondences between movement and linguistic meaning occurs. (2) No distracting microstructural associations are primed by the gesture; e.g. by similarity of the gesture’s shape to a single object appearing in the storyworld. Although occasionally subjects reported that they had started reading with searching for microstructural correspondences, this was not only extremely rare, but also apparently unsuccessful, so far as subjects reported in the debriefing. One theoretical uncertainty relates to the image-schematic modality the obtained effect rests on. Genuine motor-preparation is one possibility, yet the visual icon could also have produced the effect because of a topological correspondence between visual image and plot. The present design thus offers no real measure for the degree that genuine proprioception/kinaesthetics played a role in producing the imageschematic effects. (Many of the participants reported that a visual and not a kinaesthetic image had served as their memory cue. Yet, these selfreports do not exclude that a proprioceptive image of the movement’s “feel” was formed subliminally by cross-modal spill-over). The major problem with the results remains that only one group yielded a significant effect. One possible explanation is that some stories 24 Image Schemas in Narrative Macrostructure had too irregular and distracting features. This is not likely with twelve different stories in a group, but surely possible. Complexity, embedded sub-goals, overall story length, number of protagonists, and distracting words were largely controlled for, so that the most likely factor to bias the results is the entertainingness of stories. This is a characteristic in which stories differed to some degree, as some participants remarked. An alternative explanation is that features of the planned gestures account for the difference between the two groups. It could either be based on differences in complexity between the gestures (mono, bi- and triphasal) or based on the fact that the gesture’s final phase was dissimilar across conditions in one group but not the other. After all ordinary non-blocked FORCE vs. FORCE BLOCKAGE REMOVAL (Group 1) are non-identical in their middle phase, but both end in the same dynamical open swing.These alternatives may not be as distinct as those used in the more successful Group 2, at least for readers to whom a gesture’s end state is most salient. In view of all this, a follow-up study should seek to accomplish three things: (1) Apply a story grammar coding as a prior check of the structural similarity of the story stimuli; (2) devise a control experiment to find out whether the found effect is really due to motor planning or some other mechanism, and (3) understand better why the effect is found only for the target sentence, but not for the two others before it. Acknowledgments My thanks go to Barbara Kaup and Jana Lüdke for collaborating in the empirical part of this research. Works Cited Baddeley, A. 1986. Working memory. Oxford: Clarendon. Bartlett, F. C. 1932. Remembering: a study in experimental and social psychology. Cambridge: Cambridge University Press. Conrad, Joseph 1995 [1899/1901]. Heart of Darkness. London: Penguin. Dijk, T. A. v. 1980. Macrostructures: an interdisciplinary study of global structures in discourse, interaction, and cognition. Hillsdale, N.J.: L. Erlbaum Associates. Dijk, T. A. v., and W. Kintsch. 1983. Strategies of discourse comprehension. New York; London: Academic Press. Ferstl, E., and W. Kintsch. 1999. Learning from Text: Structural Knowledge Assessment in the Study of Discourse Comprehension. In Michael Kimmel 25 The Construction of Mental Representations During Reading, eds. H. van Oostendorp and S. Goldman, 247-277. Mahwah, NJ: Erlbaum. Freeman, D. 1993. #According to My Bond': King Lear and Re-Cognition. Language and Literature, 2(1): 1-18. —. 1995. 'Catch[ing] the Nearest Way': Macbeth and Cognitive Metaphor. Journal of Pragmatics, 24: 689-708. Gibbs, R. W. 2005. Embodiment. Cambridge: Cambridge University Press. Gibbs, R. W., and H. L. Colston. 1995. The Cognitive Psychological Reality of Image Schemas and their Transformations. Cognitive Linguistics, 6(4): 347-378. Graesser, A., C. Bowers, B. Olde, K. White, and N. Person. 1999. Who Knows What? Propagation of Knowledge Among Agents in a Literary Story World. Poetics, 26: 143-175. Graesser, A., V. Pomeroy, and S. Craig. 2002. Psychological and Computational Research on Theme Comprehension. In Thematics in Psychology and Literary Studies, eds. M. Louwerse and W. van Peer, 19-34. Amsterdam & Philadelphia: John Benjamins. Greimas, A. J. 1966. Sémantique structurale: recherche et meéthode. Paris: Larousse. Guindon, R., and W. Kintsch. 1984. Priming Macropropositions: Evidence for the Primacy of Macropropositions in the Memory for Text. Journal of Verbal Learning and Verbal Behavior, 23: 508-518. Hampe, B., ed. 2005. From Perception to Meaning: Image Schemas in Cognitive Linguistics. Berlin & New York: Mouton de Gruyter. Johnson, M. 1987. The body in the mind: the bodily basis of meaning, imagination, and reason. Chicago: University of Chicago Press. —. 1993. Moral imagination : implications of cognitive science for ethics. Chicago ; London: University of Chicago Press 1993. Kimmel, M. 2002. Metaphor, Imagery, and Culture. University of Vienna, Vienna. —. 2005. From Metaphor to the "Mental Sketchpad": Literary Macrostructure and Compound Image Schemas in Heart of Darkness. Metaphor & Symbol, 20(3): 199-238. —. (n.d.). The scope of image schemas in narrative and event structure: Towards a typology of functions. Unpublished manuscript, Vienna. Kintsch, W. 1993. Information Accretion and Reduction in Text Processing: Inferences. Discourse Processes, 16: 193-202. Kintsch, W. and J. Yarborough. 1982. The Role of Rhetorical Structure in Text Comprehension. Journal of Educational Psychology, 74: 828834. 26 Image Schemas in Narrative Macrostructure Lakoff, G. and M. Johnson. 1999. Philosophy in the flesh: the embodied mind and its challenge to Western thought. New York: Basic Books. Lakoff, G. and Z. Kövecses. 1987. The Cognitive Model of Anger Inherent in American English. In Cultural Models in Language and Thought, ed. D. a. N. Q. Holland, 195-221. Cambridge: Cambridge University Press. Langacker, R. W. 1987. Foundations of cognitive grammar. Stanford: Stanford U.P. Mandler, J. M. 1984. Stories, scripts, and scenes: aspects of schema theory. Hillsdale, N.J: L. Erlbaum Associates. —. 1992. How to build a baby: II. Conceptual primitives. PsychologicalReview, 99: 587-604. Narayanan, S. 1997. Moving right along: A computational model of metaphorical reasoning about events. University of California, Berkeley. Oakley, T. (forthcoming). Image Schema. In The Handbook of Cognitive Linguistics, D Geeraerts and H. Cuykens. New York: Oxford University Press. Parrill, F. 2000. Hand to Mouth: Linking Spontaneous Gesture and Aspect. Unpublished BA thesis, University of Chicago, Chicago. Peña Cervel, S. (2003). Topology and Cognition: What Image-schemas reveal about the Metaphorical Language of Emotions. Muenchen: LINCOM. Petterson, B. 2002. Seven Trends in Recent Thematics and a Case Study. In Thematics: Interdisciplinary Studies, eds. M. Louwerse and W. van Peer, 237-252. Amsterdam & Philadelphia: John Benjamins. Spivey, M., D. Richardson, and M. Gonzalez-Marquez. 2005. On the spatial and image-schematic underpinnings of real-time language processing. In Grounding Cognition: The Role of Perception and Action in Memory, Language, and Thinking, eds. R. A. Zwaan and D. Pecher. Cambridge: Cambridge University Press. Stern, D. N. 1985. The interpersonal world of the infant: a view from psychoanalysis and developmental psychology. New York: Basic Books. Stockwell, P. 2002. Cognitive poetics: an introduction. London; New York: Routledge. Sweetser, E. 1990. From etymology to pragmatics: metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Talmy, L. 2000. Toward a cognitive semantics. Cambridge, Mass.; London: MIT Press. Michael Kimmel 27 Turner, M. 1996. The literary mind. New York: Oxford University Press. Vipond, D. 1980. Micro- and Macroprocesses in Text Comprehension. Journal of Verbal Learning and Verbal Behavior, 19: 276-296. Zacks, J. and B. Tversky. 2001. Event Structure in Perception and Cognition. Psychological Bulletin, 127(1): 3-21. Zwaan, R., J. Magliano, and A. Graesser. 1995. Dimensions of Situation Model Construction in Narrative Comprehension. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21: 386397. SHORT BIO