| 37

Constructing Domains in Visual Narratives:

Structural Patterns of Incongruity Resolution
BIEN KLOMBERG, JOOST SCHILPEROORD, NEIL COHN

Abstract: Understanding visual narrative sequences, like those in comics, requires readers to track
situational information like the continuity of characters, locations, or events across panels. Yet, some
sequences intentionally present incongruities, which may elicit the inference that the narrative
presents two separate domains: an established, expected storyworld (primary domain) and an addi-
tional context surrounding the incongruous events (auxiliary domain), like a character’s imagina-
tion, dreams, or memories. This paper describes how these inferences are supported by visual con-
structional patterns, which emerge across a wide range of narratives, yielding further insight into the
fundamental role of graphic and structural cues within visual storytelling.
Keywords: incongruity resolution, domains, parallel architecture, visual language, comics.

T o understand visual storytelling, readers need to track situational information like characters,
places, and events across panels, and integrate these coherently in a mental representation of the
narrative (Cohn, “Your Brain on Comics” 355; Dijk and Kintsch 5; Huff et al. 943; Loschky et al.
312). Generally, information that is discontinuous with the current context is a cue for an event
boundary (Cohn, “Your Brain on Comics” 360; Gernsbacher, “Coherence Cues” 3; Loschky et al.
322), but some sequences may include intentionally incongruous information that needs to be mean-
ingfully incorporated in the current storyline. How do we deal with such issues of incongruity?
Consider Figure 1, which shows short comics from JA!, a series about the experiences of its two
collaborative creators (who appear as characters in the final scenes of each strip). In all the strips in
Figure 1, the final panel is incongruous with the prior series of events, juxtaposing distinct places and
agents within a layout. This divergent information typically prompts readers to engage in “incon-
gruity resolution”, i.e., finding an interpretation that explains the incoherence (Schilperoord 2–4,
18–19; Forabosco 47). For Figure 1, readers can resolve the differences across characters/locations
when inferring that the first series of events are not actually occurring or present at that moment/
place, but function as metaphoric depictions of the “real” characters’ actions and/or feelings. In
Figure 1a, the mole’s actions are analogous to the woman’s: she bumps into things while “as blind as
a mole”, seeking the glasses worn by the dog. Similarly, Figure 1b’s piranha (devouring meat quickly)
and snail (eating a leaf slowly) seem to reflect the eating habits of the man and woman, while Figure
1c compares the woman catching a spider for the man to a superhero defeating a monster to save a
civilian. Here, the animals, superhero, and monster would not be physically present in the focal
characters’ storyworld but can be inferred as additional contexts applied to those characters. The
recognition of these incongruities and their resolutions are essential to the humor intended by these
comics (Forabosco 47), and give a reason for why this incoherence was shown (Schilperoord 4): they
reflect the characters’ inner thoughts and feelings.

Journal of Comparative Literature and Aesthetics Vol. 47, No. 3, Autumn 2024 [37-55]
© 2024 Vishvanatha Kaviraja Institute, India
38 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

Figure 1. Examples of incongruous events that can be meaningfully incorporated.

All sequences are from the series JA! © Ángela Cuéllar and Jonás Aguilar (Creative Commons License).

Intentional incongruities like in Figure 1 have been widely recognized in visual narratives, united
by the inference that some events do not actually occur at that place/moment. This gives rise to two
contexts that should be integrated during incongruity resolution: a primary domain (the established
storyworld, e.g. the man and woman) and an auxiliary domain (an additional context associated with
events inferred to not occur, e.g. the analogy with animal habits/superheroes). Such inferencing may
make sequences more complex but also engaging, rewarding, and fun, since readers are challenged
but can still manage to construe meaning (Ramachandran and Hirstein 31; Giora et al. 138). In
Figure 1, these domains (and their resolution) provide the strips with their aesthetic appeal. Such
inferential challenges may also heighten narrative tension, e.g. for horror and mystery genres that
may deliberately prolong cues that help to resolve incongruity. Since the resolution of these se-
quences relies on successfully integrating two domains, and such phenomena appear to rely on
patterned structures, we refer to these as domain constructions, but similar phenomena have been
subsumed by work on perspective taking or focalization (Maier and Bimpikou; Mikkonen 110;
Horstkotte and Pedri 38–40).
Although prior approaches have attempted to describe the semantics related to resolving the
incongruities within domain constructions (Abusch and Rooth 8; Bimpikou 246; Maier and
Bimpikou), they have not addressed what visual forms might cue domain constructions, and how
these cues correspond to meaning or the structural patterns that arise across such features of form. In
Figure 1, the layout of the three sequences is such a form-related cue: all the panels showing the
animals’/superhero’s actions create a grouping, while the panel showing the human characters is its
own row. These groupings are reinforced by panel borders, with the first four/six panels using
explicit borders, while the final panel has none. Additionally, aspects of form may create connections
between primary and auxiliary figures. In Figure 1a-b, the eyes of the mole/snail are drawn the same
way as the woman’s eyes, and in Figure 1b and 1c, the red of the meat/ civilian’s shirt corresponds to
the man’s shirt, while the green of the leaf/superhero’s top corresponds to the woman’s shirt, and the
superhero mask in 1c to her red glasses. Overall, formal features align with the semantic incongruity
and may cue intended relations between domains and their respective figures.
This paper aims to account for such patterns and how they may signal incongruity to be resolved.
Our examples include a variety of comics by different authors, highlighting how domain construc-
tions appear widespread in storytelling for authors to convey characters’ personal experiences, while
aligning these meaning relations with conventionalized graphic features. The analyses are embed-
 Constructing Domains in Visual Narratives | 39

ded in a cognitive and linguistic approach to visual representations, and as such, aim to describe the
patterned nature of these encoded constructions. We thus investigate the effects of graphic and
structural cues on (inferential) meaning-making in storytelling and what constructional patterns
persist across authors and their works. Additionally, such patterns could apply to other visual media,
e.g., film, which maintain many of the same affordances as drawn visual narratives. Naturally,
creators may also rely on similar schemas when working across media, often because film storytelling
first occurs as drawn storyboards. To discuss domain constructions for comics, we first present how
visual narratives are processed and how previous work has described issues of domain constructions
similar to Figure 1. Next, we propose a model that describes patterned correspondences between
form and meaning related to this construction.

Domain constructions in visual narratives

To comprehend visual sequences in general, readers integrate story components into one coher-
ent, mental representation of the narrative (Cohn, “Your Brain on Comics” 355; Dijk and Kintsch
5; Huff et al. 943; Loschky et al. 312). This representation is known as a situation model, and includes
situational aspects such as characters, their motivations, spatial locations, time, etc. (Dijk and Kintsch
46; Zwaan and Radvansky 167; Gernsbacher, “Coherence Cues” 5). With each incoming piece of
information, the situation model gets updated, and sufficiently congruent elements can be mapped
onto the existing model (Cohn, “Your Brain on Comics” 360; Gernsbacher, “Coherence Cues” 3;
Loschky et al. 321). The more incongruent the incoming information, the greater the updating costs
(Huff et al.; Hutson et al.; Magliano et al.), until significant discontinuity sponsors an event boundary
(Gernsbacher, “Two Decades” 5; Loschky et al. 322).
While sequences with auxilairy domains are incongruous and have been shown to require more
updating indeed, these were equally comprehensible as sequences with only a primary storyworld
(Klomberg, Fadeeva, et al.). Moreover, domain constructions seem to be fairly common, with cor-
pus work showing they are prevalent in around 65% of comics (van der Gouw et al. 24). The
ubiquity and comprehensiblity of domain constructions give rise to the question of what mecha-
nisms underlie this storytelling technique.
Prior work converges on a general semantic distinction between visual events that are perceptible
to all characters versus those that are perceptible only to a single character (Abusch and Rooth 8;
Bimpikou 247; Duncan 275; Horstkotte and Pedri 13, 147–48; Maier and Bimpikou). To mark this
distinction, we use the notion of domains (Clausner and Croft 2; Langacker 147), which denote
particular types of mental spaces (Fauconnier and Turner 40) within the context of narrative se-
quencing, and thus a certain context that visual events can be attributed to. The primary domain
represents events that are inferred to transpire in the “actual” storyworld, i.e., the world physically
accessible or perceivable to all characters. An auxiliary domain on the other hand, represents events
inferred as “not physically occurring at that time”. Auxiliary domains are typically resolved as
depictions of a (private) mental experience of a certain character, such as their fantasy, hallucina-
tions, dreams, or memories (Horstkotte and Pedri 40–48; Abusch and Rooth 12; Maier and Bimpikou;
Bimpikou 255; Mikkonen 110). Work on focalization (Horstkotte and Pedri 1) describes domain
constructions by analyzing how a narrative may present the experiences of a certain character, while
approaches from semantics and discourse theory focus on the veridicity of events (Bimpikou 247;
Maier and Bimpikou).
In addition to semantic distinctions between events, prior work foregrounds the role of perspec-
tive-taking for these sequences (Abusch and Rooth 8; Bimpikou 249; Maier and Bimpikou), with
inferences of subjectivity as cue for auxiliary domain events. Consider the modified example in
Figure 2a, exemplifying a two-panel construction that implies subjectivity. When a panel with a
character looking off-panel (the man in red) is followed by an object incongruous with the narrative
(the snail), the sense of perspective-taking evoked by this two-panel construction facilitates the
40 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

inference that the incongruity is a perception private to the character, i.e. their imagination, dream,
or hallucination (Abusch and Rooth 8; Bimpikou 248–49; Maier and Bimpikou). This contrasts
with “blended-perspective” (Bimpikou 246; Maier and Bimpikou), where elements of distinct do-
mains are merged in a single panel. In Figure 2b, a man looks into the window of a shop but sees a
reflection of a different person, with no apparent co-reference. Contextually, only the man can see
the reflection of this dark-haired figure, who also haunts him in his dreams later on. Thus, this
reflection seems to be imagined by the man, as if he and this figure were one in this moment
(reinforced by the comic’s title: Bonding). In this way, blended-perspectives merge the physical
storyworld (the shopping street) with the private thoughts of a character (the reflection of the dark-
haired figure).

Figure 2. Examples illustrating a) a two-panel construction and b) a blended perspective. Figure a is a

modified version of a sequence from the series JA! © Ángela Cuéllar and Jonás Aguilar
and Figure b from Bonding © 2015 Lumarin and Marina Privalova.

While the availability of an identifiable experiencer may be helpful for the recognition of domain
constructions, not all auxiliary domains can be readily interpreted as a particular character’s experi-
ence. For Figure 1a-c, it is unclear whether these sequences are imagined by a storyworld character
(does the woman imagine herself as a mole/snail/hero?) or a comparison made by a non-diegetic
narrator. Despite this conceptual ambiguity of why the two event types are integrated together, the
cues sponsoring those inferences (e.g. layout, visual patterns, colour correspondences, etc.) can be
identified without issues. It thus seems that these cues can function independently from meaning.
These and other patterns have received little focus across research on visual narratives, despite
some prior observations of cues for individual images. For instance, previous analyses acknowledge
that divergent panel borders or changes in layout correspond to incongruous events (Duncan 275;
Horstkotte and Pedri 40–46; Mikkonen 110–11), but they have not been generalized beyond the
specific examples that are addressed. Moreover, there are patterns not yet accounted for, e.g. Figure
1’s segmentation of panels and its ambiguous (or lacking) experiencer. Thus, while existing theories
have yielded valuable insights into the meaning-making processes involved in domain resolution,
these do not acccount for the formal patterns that signal these construals, which may give the false
impression that such construals appear without patterned contexts. How then can we account for
these various visual narrative constructions?
 Constructing Domains in Visual Narratives | 41

The Domain Constructions model

To account for the structured patterns of form and meaning that occur in domain constructions,
we here propose the Domain Constructions model, or DC model. The model describes structural
correspondences between form and meaning that occur for domain constructions, where incongru-
ity across situational aspects and form-related features cue the search for a meaningful interpreta-
tion. This involves integrating an auxiliary domain (events inferred to not physically occur at that
time/place) within the story’s primary domain (the physical storyworld accessible to all characters).
The proposed model is not restricted to cues signalling domains across panels but also pertains to
elements within panels (Bimpikou 245–46; Horstkotte and Pedri 40–42; Maier and Bimpikou).
The DC model is embedded within the broader framework of Visual Language Theory (Cohn,
The Visual Language 1–13) which posits that graphic communication maintains structures and cog-
nition consistent with natural languages, as has been supported both theoretically and empirically (see
Cohn, Who Understands Comics? (20–22)). Visual Language Theory itself is embedded within the
linguistic model of the Parallel Architecture (Jackendoff 107–26), which characterizes language in
terms of three parallel, but independent structural components: modality (phonology), grammar (syn-
tax), and meaning (conceptual structure), as well as interfaces that emerge between these. The Paral-
lel Architecture has been applied to characterize visual language in terms of the same components of
modality (graphic structure), grammar (combinatorial structure), and meaning (conceptual struc-
ture) (Cohn, Who Understands Comics? 5–6). These three structures, shown in Figure 3 below,
describe both single units, e.g. a single panel, and sequences of units, e.g. multiple panels on a page.

Figure 3. The Parallel Architecture framework.

In visual expressions, each component involves unit- and sequence-level structures, indicated by
the vertical arrows in Figure 3. The modality component (i.e., graphic structure) related to single
panels is termed graphological structure (Klomberg, Hacimusaoglu,
Hacımusaoğlu et al. 10), which characterizes
the organization of the visual markings that make up the images, such as lines, dots, or colours, and
the visual regions they compose. Graphic structure across panels is captured by layout, which ac-
counts for how sequences of images are integrated spatially on a canvas. The grammar component
characterizes the combinatorial structure of a single panel in terms of its morphological structure,
and organizes sequences of panels according to principles of Narrative Grammar (Cohn, Who
Understands Comics? 12–16). Finally, the meaning component describes how panels are understood
42 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

independently (conceptual structure) and in sequence (event structure). For domain constructions
such comprehension relies on similar principles, as the same conceptual inferences apply within and
across single panels.
While domain constructions would be resolved at the level of conceptual structure, the DC model
accounts for the ways that graphic and combinatorial structures contribute significantly to their
recognition and resolution; these interfaces across components are expressed by the horizontal arrrows
in Figure 3. Hence, the following sections will detail how constructions encoded across these com-
ponents signal auxiliary domain usage and/or support relations of co-reference across entities that
initially appear non-co-referential. We first present structural correspondences between the graphic
and conceptual structures, then between grammatical and conceptual structures, and finally, struc-
tural correspondences across all three components.

Graphic Structure
Graphological Structure. To describe how form may interact with meaning we must first
characterize the graphic form of visual representations. Analogous to how phonological structures
organize sound structures, a graphological structure (Klomberg, Hacımusaoğlu
Hacimusaoglu, et al. 10) organizes
visual-graphic components. The basic units of those components are graphemes, i.e. dots, lines,
curves, and spirals, which combine to create larger structures of “regions”: enclosed visual groupings
(either explicitly so or inferred) (11). Regions, in turn, can recursively combine to form even larger
graphic structures that correspond with conceptual representations, i.e., they can be recognized as
meaningful objects (Willats 94). Graphological structure is thus concerned with purely graphic
information, such as the lines and proporties of regions. We here propose that differences across
these graphic features are often used to signal domain constructions.

Figure 4. Examples illustrating graphological cues corresponding to meaning.

Figure a is from Socrate © Francesco Barilli and Alessandro Ranghiasci,
Figure b from Jar of Fools © Jason Lutes, and Figure c from Raruurien © Ann Maulina.
 Constructing Domains in Visual Narratives | 43

Figure 4 illustrates three types of graphicological cues that may signal different conceptual do-
mains: colour, linework, and shape similarity (Horstkotte and Pedri 12; Mikkonen 116–20). Consider
Figure 4a, where the first page shows the philosopher Socrates during his own trial, the second page
presents Socrates amongst different people in a different location, and the third page shows different
figures in an unknown environment. These differences in both characters and locations confront the
comprehender with an incongruity. Crucially, each page here is marked with a distinct colour scheme.
The incongruity across pages can be resolved by construing the second page’s events as taking place at
another moment (Socrates teaching his students), with the same agent in both scenes (Socrates). The
third page can subsequently be construed as the contents of Socrates’ lesson. These inferences are
signalled by the pages’ graphological differences, with the first page’s trial (the primary domain)
shown in colour, the second page’s memories (an auxiliary domain) in black and white, and the third
page’s exposition (a second auxiliary domain) in black and orange hues, like a Greek vase (Andrew
et al. 2). A similar alignment of colour schemes and domains occurs in Figure 4c, where the final
incongrous panel can also be inferred as a past event featuring the same agent (the woman). This
panel is coloured only in blue hues, as opposed to the range of colours used in previous panels.
Differences in the linework of drawing styles may also correspond to domain constructions. In
Figure 4a, the third page’s drawing style further mimics Greek vase art , with figures shown mostly
from the side and without shading (Andrew et al. 5). This stylistic shift again aligns with the concep-
tual interpretations that arise for page 2 and 3. While both pages show colour differences that
correspond to different domains, page 2 maintains the same drawing style as the one used for
primary domain events on page 1, warranting the interpretation that the memory event bears a
substantive conceptual relation to Socrate himself, i.e. his lived experience. In contrast, page 3’s
features elicit associations to the myths and legends typically shown on those Greek vases (Andrew et
al. 12), which aligns with the idea that this event is fictitious even to Socrates himself. Seemingly, the
use of differing drawing styles across pages 2 and 3 in Figure 4a functions to distinguish between the
two different auxiliary domains.
Graphological differences signalling different domains also emerge from the linework of the panel
borders. In Figures 4a-c, diffferent panel borders may support inferring those panels as correspond-
ing to auxiliary domains. Figure 4a’s third page has thick, decorated borders which contrast the
previous two pages’ borders, thus emphasizing that this last page may present a second auxiliary
domain rather than a continuation of the memory on page 2. In Figure 4b, the use of different
borders likewise supports a distinction between the events captured in those panels: the man dreams
that he is dancing with the woman (shown in panels with rounded borders), and wakes up to find he
is actually sitting in a park (shown in panels with square borders). In Figure 4c the final panel has
distinctly frayed borders, compared to the previous rectangular borders, reinforcing that the final
event is part of an auxiliary domain (here, a memory). Besides these differences in panel borders’
linework, differences can also arise when borders are omitted, as in Figure 1. Overall, incongruity in
lines across panels may signal that additional updating is needed.
Finally, similarities in shapes across panels can motivate different domains. In Figure 4b, the switch
between domains occurs between panels 8 and 9, which both depict the main character. Noticeably,
the character’s composition is highly similar across panels, with his hands in a similar position
relative to his face despite changes in his expression, hair, clothing, and background. The overall
similarity alongside small differences seems to emphasize that this same moment occurred across two
domains (him holding up his hands in a dream and in the physical storyworld). Experimental re-
search (Klomberg, Fadeeva, et al.) suggests that familarity with comics may mediate whether such
contour similarity affects processing, with more experienced readers slowing down for sequences
lacking this similarity.
As shown by our analyses, these graphological cues do not indicate the presence of auxiliary/
primary domains themselves, but rather do so in contrast to other panels (Mikkonen 119). For
44 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

example, a whole sequence can have rounded, wavy, or no borders or be shown in greyscale or blue
hues, but this alone does not signify auxiliary events. When these features contrast other panels that
have been established as normative to a primary domain, it may prompt readers to infer the events
with incongruous features as part of an auxiliary domain. These graphological patterns thus include
open slots, filled with whatever applies to a specific comic (e.g. Figure 4b has rounded borders as
incongruent vs. square borders as expected, while Figure 4c contrasts wavy vs. square borders). Such
cues often apply to several panels; e.g., Figure 4a-b’s colours schemes and divergent borders persist
for all panels depicting those auxiliary domain events. Additionally, graphological features can easily
combine, whether border differences and shape similarity (Figure 4b), colour and border differ-
ences (Figure 4c), or colour, style, and border differences (Figure 4a).
Layout. Layout refers to the part of the graphic structure that concerns the spatial organization of
panels on a larger canvas (e.g. a page) (Cohn, Who Understands Comics? 9–11). We already de-
scribed one pattern of layout for Figure 1, where the auxiliary domain panels cluster together
followed by a single panel showing the primary domain. This general construction is shown in
Figure 5a where tree structures are used to specify the organization of these sequences.

Figure 5. Examples illustrating layout trees with the * indicating the introduction of a second domain
(whether primary (b) or auxiliary (c)). Figure b is slightly adapted from the series JA! © Ángela Cuéllar and
Jonás Aguilar, and Figure c is slightly adapted from REAL © Inoue Takehiko.
 Constructing Domains in Visual Narratives | 45

Figure 5b shows how Figure 1’s examples would fit this organization: the events in one domain
(panels 1 to 4) are structured as multiple horizontal rows stacked on one another vertically. This
column of panels then combines with another row underneath that corresponds to the second do-
main (panel 5), forming an ultimately vertical layout. This construction can have additional internal
structure within its lower-level vertical grouping; e.g., Figures 1a and 1c feature two rows, while
Figure 1b features three. Note that Figure 5b is ambiguous: it could also be viewed as three horizon-
tal rows stacked upon each other, rather than a column upon a row. Nevertheless, either interpreta-
tion maintains the general construction illustrated in Figure 5a and demonstrates a recurrent pattern
of segmentation across panels that aligns with conceptual inferences, with a sequence’s second do-
main corresponding to a grouping on its own.
A second instantiation of the general pattern is shown in Figure 5c, where the horizontal row
combines with other horizontal structures not only before but also after. Here, a boy in a hallway
(panels 1-2) remembers himself and a girl on a motorbike (panel 3). This panel on its own (panel 3)
corresponds to an auxiliary domain of a past event. In Figure 4a, different domains are distributed on
different pages. Thus, it seems that layout segmentation may support conceptual interpretations of
domains by presenting distinct panel groupings within page structures as well as across pages. Such
segmentation echoes other observed alignments between coherence shifts and layout structure
(Hacimusaoglu
Hacımusaoğlu et al.).
All the patterns discussed so far strictly concern the organization of panels and/or pages, and render
available slots that can be filled by particular forms, rather than specific meaning. In Figure 5b, the
single panel functioning as a row corresponds conceptually to a primary domain, while in 5c, the
single panel corresponds to an inference of an auxiliary domain. This emphasizes that these con-
structions are aspects of layout which can interface with meaning in various ways. These patterns
describe how the narrative switches domains in general, regardless of the direction of this switch, or
the particular nature of the domains involved (Klomberg, Fadeeva, et al.).
Nevertheless, one layout structure appears to relate to auxiliary events more directly. Figure 6
shows two collage layouts, where multiple scenes graphically overlap without overt panel borders.
Here, both pages show events that can be inferred to have taken place in the past and can thus be
identified as auxiliary domain events. Collage layouts potentially emphasizes the sense that such
events are “flashes” of memory. Corpus work indeed found that 83.8% of collage framing contained
auxiliary domains (van der Gouw et al. 36), supporting a close relationship between them.

Figure 6. Examples illustrating collage sequences. Figures are from Valley of White Birds
© Wolfsmoke studios and from Pirozhki © Sideburn004, respectively.
46 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

Combinatorial Structure
Morphological Structure. This component of the Parallel Architecture is concerned with the
patterns of abstracted forms of visual units (Cohn, The Visual Language 23–24; Cohn, “Combinato-
rial Morphology”; Cohn, Who Understands Comics? 12). Similar to morphology in verbal language,
visual morphology contains units of visual language that can stand on their own (e.g. a single char-
acter or object) next to bound morphemes that need to combine with units to be meaningful. These
components use various strategies to combine, such as via suppletion, e.g. replacing a character’s
eyes with hearts to show love, or blending, e.g. depicting a character’s normal body with an animal’s
head (Cohn, The Visual Language 44–47; Cohn, “Combinatorial Morphology”).
These morphological patterns may signal domain switches and semantic (co-referential) connec-
tions across characters from different domains. Figure 7a illustrates a gradual replacement (partial
suppletion), where the man’s increasingly disturbing facial features from panel 4 to 7 signal that the
woman slowly realizes his malicious intentions. Figure 7b’s morphology signals the start of a dream:
a man was watching a movie with a female protagonist (panels 1-3), but evidently fell asleep as the
movie becomes personal and nightmarish, and he wakes up on a later page. He apparently falls asleep
around panel 4, when the actress’ appearance blends with his own. This “morphological blend”
shows parts of one meaningful entity (the woman’s hair, clothes, and earrings) merged with parts
from another entity (the man’s own face and body hair), which supports the inference that this is a
dream rather than a movie. Finally, Figure 7c shows visual components repeated across different
figures, with the woman and animals having the same eyes and eyebrows. These we term “echoic
morphemes”: visual components that mimic one another morphologically. Rather than signalling a
domain switch, repeating visual components may signal readers to infer some semantic relation
between characters, and as such, may support subsequent inferences of co-reference across these
units (Klomberg, Hacımusaoğlu
Hacimusaoglu, et al. 21).

Figure 7. Examples illustrating the morphological strategies of a) partial suppletion, b) morphological

blend, and c) echoic morphemes. Figure a is slightly adapted from Vicki Lante © Patrick Steptoe, Figure b is
slightly adapted from Jar of Fools © Jason Lutes, and Figure c is slightly adapted from JA!
© 2016 Ángela Cúellar and Jonas Aguilar.
 Constructing Domains in Visual Narratives | 47

Narrative Structure. Where morphology is concerned with the relations between visual units
within panels, narrative structure characterizes the combinations of panel units, along with the
function these panel units play within a narrative sequence. The combinatorial structure of se-
quences is captured by Visual Narrative Grammar (VNG) (Cohn, “Visual Narrative Structure”
420–38; Cohn, Who Understands Comics? 43–46), which identifies the narrative categories of panels
and the hierarchic embedding of larger sequences of panels. We first explain the narrative categories
panels can have, and then analyze how these categories may interface with domain switches.
Figure 8 shows the narrative structure of the comics in Figure 1 according to VNG (Cohn, “Visual
Narrative Structure” 420–27; Cohn, Who Understands Comics? 43–46). Let’s start with the first
grouping in each of these sequences. In Figure 8a, panel 1 introduces the scene and characters (an
Establisher), and then panel 2 sets up (an Initial) the climactic moment in panel 3 (the Peak), until this
tension diminishes in panel 4 (a Release). These four panels follow the canonical narrative schematic
order (E-I-P-R) within VNG. Figures 8b and 8c use an additional “conjunction schema” where
identical categories follow one another. Figure 8b uses an action-conjunction (see subscript a),
where panels of the same narrative categories repeat similar actions. Figure 8c shows an environ-
mental-conjunction (subscript e), where the depicted characters or objects are inferred to belong to
the same environment.
Narrative categories not only apply to individual panels, but also to constituents of panels. In
Figure 8a, the first four panels constitute the climax of the sequence, thus forming a Peak sequence,
while the final panel diminishes that tension and acts as a Release for all preceding panels. Figure 8b
and 8c use this same pattern, with a final Release panel following a Peak segment.

Figure 8. Examples illustrating grammatical tree structures with a a) canonical scheme, b) action conjunc-
tion, and c) environmental conjunction. Figures are slightly adapted sequences from JA!
© 2016 Ángela Cúellar and Jonas Aguilar.
48 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

For these sequences, narrative constituents correspond with the conceptual division between
domains. Note that this division persists despite internal variation within the constituents, with the
Peak-Release sequence on the highest level corresponding to a division between auxiliary and
primary domain events. Other narrative constructions, e.g. conjunctions, could similarly support
domain constructions across panels, and will be investigated more extensively in future works. As
shown for other structures, these patterns can be expected to occur regardless of what conceptual
domain fills what slot.

Meaning
Conceptual Structure. Conceptual structure specifies how representations are involved in mean-
ing-making, which in the case of domain constructions often involves complex inferencing. Incon-
gruity occurs across situational information, with time, space, or characters mismatching across panels,
signalled also through differences in graphology, layout, morphology, and grammar. Such incongru-
ity can then prompt the resolution that those depictions reflect events that do not actually occur in the
storyworld (the primary domain), but instead correspond to some alternative context (an auxiliary
domain). All examples shown so far involved such inferencing. Despite deviating from the primary
domain, resolving these incongruities allows for coherent integration into situation models.
The way that domains are to be integrated to arrive at a sensible resolution will depend on the
inferrred nature of the auxiliary domain, i.e., imagination, hallucinations, dreams, and memories
(Abusch and Rooth 12; Bimpikou 255; Horstkotte and Pedri 40–48; Mikkonen 110–12). For instance,
incongruent appearances of a character can be resolved as happening in a dream in which the character’s
appearance transforms, rather than this character actually (physically) changing or that a new charac-
ter is being introduced. Particular graphic and/or structural patterns may be used for certain types of
auxiliary domains; e.g., greyscale panels seem more indicative of memories than hallucinations, for
which we may instead expect bright or atypical colours. So far, the DC model categorizes nine types
of auxiliary domains: Imagination, Knowledge, Dream, Memory, Foresight, Hypothetical, Fiction,
Fourth Wall Break, and Analogy, along with cases that may remain Ambiguous.
Imagination and Knowledge domains are inferred as events that occur in the mind of a single
experiencer during an active (awake) state. The page in Figure 9a (read from right to left) precedes
the “blended-perspective” from Figure 2b, and implies the reflection is imagined: the man with the
cap (the experiencer) sees an incongruent reflection in the shop window (panel 3), which only he
appears (able) to see, as the reflection disappears when someone else is present (panel 6). Generally,
Imagination domains show fantasized events originating from a single, awake character that other
characters seem unable to perceive, e.g. active imagination, daydreams, or hallucinations. In Figure
9b, the swimming techniques can also be inferred to occur in the mind of the boy only, due to the
juxtaposition of these techniques and the close up of the boy’s face, and reinforced by the absence of
physical materials he could be looking at in the rest of the sequence. Rather than a fantasized event,
however, these seem a visual representation of the boy’s knowledge relating to specific techniques.
While both domains appear private to one experiencer, Imagination typically depicts made up
information, while Knowledge reflects thoughts of a more substantive nature, such as spatial knowl-
edge (e.g. a map of the environment), encyclopedic knowledge (e.g. how a certain object looks or
functions), or instructions (e.g. how to perform certain actions).
While these events occur during characters’ awake state, experiencers may also visualize events
involuntarily, e.g. when dreaming. In Figure 9c, the third panel shows the experiencer jolting up-
right with a scream, suggesting he awakens, which is emphasized by the sixth panel depicting him on
his bed. The incongruity between the first two panels and the rest can thus be resolved by inferrring
the first two panels to represent a dream. Typically, Dream domains are preceded by a character
lying down or closing their eyes and/or followed by a character wakening or opening their eyes,
providing contextual cues that signal the incongruous events as part of a dream.
 Constructing Domains in Visual Narratives | 49

Figure 9. Examples showing an a) Imagination domain, b) Knowledge domain, and c) Dream domain.
Figure a is from Bonding © 2015 Lumarin and Marina Privalova, Figure b from
Poisson © Nie Jun, and Figure c from Jar of Fools © Jason Lutes.

These three domain types concern events assumed to originate completely from someone’s mind,
but two other types -Memory and Foresight- denote events that should be inferred to occur at some
past or future point in the story’s timeline. Figure 4a (Socrates’s trial), Figure 4c (the woman remem-
bering being angry), and Figure 6 (the collage examples) all show examples where incongruous
events have to be construed as a protagonist’s memories. Similarly, the collage panel in Figure 10a
shows the protagonist overlayed with a series of events that have occurred before, some of which
even have been shown in the story explicitly. In these instances of Memory domains, the primary
assumption is that the depicted moment is something that the experiencer recalls from the past
(which may or may not have been shown earlier in the story). Narratives can also show events that
will occur in the future, e.g. via a prophecy or a character having a vision. Figure 10b shows an
example of a witch watching her crystal ball with the vision illustrated on the right (a princess
picking up a sword, coaxed by the snake on her shoulders). This exemplifies a Foresight auxiliary
domain, representing (likely) future events.
50 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

Figure 10. Examples showing a a) Memory domain, b) Foresight domain, and c) Hypothetical domain.
Figure a and b from Valley of White Birds © Wolfsmoke studios and Figure c © Sharitha van der Gouw.

In addition to past and future events, a depicted event may or may not come to pass, i.e., a
Hypothetical auxiliary domain. Examples are thought experiments (an answer to the question “What
if…?”) or outcomes related to very specific circumstances. In Figure 9c the first panel shows an old
man saying that “the white goddess” (i.e. the white creature in the second image) may be caught off
guard only by the protagonist in panel 2 and 3, and so he “might just be the only one who could kill
her”. The panel in between these speech balloons shows such a scene where the protagonist has killed
the white goddess. However, the reader knows this event has not (yet) happened, and so the panel
shows an Hypothetical situation (cued predominantly by the modal keyword “might”), without
indicating whether the event is likely to occur or not.
Thus far, we discussed auxiliary domains representing events revolving around the experiencer
that visualizes, remembers, or otherwise inspires that event. However, auxiliary domains can also
represent events that do not appear private to an experiencer. The first three panels of Figure 11a
show different moments of a fight, while the final panel shows a person holding a comic book
showing these exact three panels. Here, the first three panels depict fictitous events (a comic book
narrative), and the final panel depicts the primary storyworld domain, where the story will continue.
This auxiliary sequence is thus part of a Fiction domain, which can be described as a ‘story within a
story’ and/or events that are fictitious to the characters themselves. Examples include depictions of
folklore history, fairytales, or other accounts of (past) events that do not involve the narrator itself but
other (fictitious) people.
 Constructing Domains in Visual Narratives | 51

Figure 11. Examples showing a a) Fiction domain and b) Fourth Wall Break domain.
Figure a is from El Verdagazo © Jorge E. Pérez and Francisco Alberto López
and Figure b from Cottonstar © Ben Geldenhuys and Danelle Malan.

Additionally, narratives may also suspend the established storyworld altogether, as in Figure 11b
where the first panel shows the comic creators reacting to a story event on the previous page, while
the second panel continues in the primary storyworld domain. Thus, this first panel steps out of the
story for a moment and visualizes figures with no relation to the narrative at all (except supposedly
having created it). The incongruity across figures and places is resolved when the first panel is
interpreted as an event from the creator’s own world. We summarize these as instances of a Fourth
Wall Break domain. It is worth pointing out that this also allows visual narratives to be self-referen-
tial, and to have creators comment on their own creation.
The last type of auxiliary domain is Analogy, which resolves incongruity via a (metaphoric)
comparison between two domains. This was the case for Figure 1, where the animals and superhero
figures metaphorically represented the actions and/or feelings of the actual characters shown in the
final panel. For these domain constructions, it may be ambiguous whether an diegetic experiencer is
behind the analogy (i.e., whether the woman imagines herself as a mole/snail/hero) or whether the
comparison can be attributed to a non-diegetic narrator/author.
Finally, auxiliary domain events cannot always be easily categorized and may evoke multiple
possible interpretations. A sequence may not imply an experiencer explicitly, e.g. Figure 1, or Figure
10c, where the Hypothetical situation of the protagonist killing the white goddess could be attrib-
uted to the mind’s eye of either the protagonist or his mentor (or both). Even sequences with overt
experiencers may be Ambiguous, when it is unclear what action the experiencer is involved in
(imagining, remembering, etc.). In those cases, contextual cues then do not distinguish whether the
depicted scene is private to the experiencer or not, or whether the depicted scene could have already
occurred in the past. Alternatively, cues may overlap, such as when a characters’s dream features a
52 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

memory or likely future event. This ambiguity is likely presented on purpose to leave readers in
suspense or create more possible inferences (e.g. is it a dream or a prophetic vision?), requiring (or
granting) more reader involvement and freedom to interpret sequences according to readers’ own,
strongest associations, which may contribute to the pleasure of reading.

Meaning across modalities

The discussion so far related to visual information; however, discontinuities also persist in
multimodal interactions between text and images. Various examples illustrate such an interaction. In
Figures 4c, 6a, and 9d, the text includes past tense and deictic indicators (4c: “that time”, 6a: “back
then”, 10a: “my life like this…has been”), supporting the interpretation that the visual image depicts
a past event. Figure 10c is even more overt, with the modal verb “might” reinforcing the inference
of a Hypothetical domain. In Figure 11b too, the text in the first panel identifies who is depicted (“the
creators of this comic”), which would have been difficult to establish otherwise. These references
may even work in hindsight or advance, e.g. when auxiliary events are referenced in later panels
(e.g., “I had a bad dream”), textully establishing an event as a particular domain type (here, Dream).

Interfaces across Parallel Architecture structures

So far, we have discussed how the DC model operates on correspondences between graphic
structure and meaning, and between combinatorial structure and meaning, and how both support
understanding of domain constructions in visual narratives. Correspondences across these structures
may align in various ways for visual narratives, which gives rise to different ways that auxiliary
domains are presented within a sequence.

Figure 12. Illustration of the three domain switches: a Sudden transition,

a Gradual transition, and a Partial transition.
 Constructing Domains in Visual Narratives | 53

Figure 12 summarizes how narratives may switch between their multiple domains. With a Sud-
den transition, layout constructions align with grammatical constructions, such that the narrative
switches domains from one individual panel to another. In the example below, the narrative switches
to a memory event across panels 3 and 4, with each domain reflected in its own panel. This type of
transition would also apply to the two-panel constructions discussed by Abusch and Rooth (8). For
Gradual transitions, the domain switch requires more than two panels. The example below illustrates
how change across panels occurs progressively, with each subsequent panel amplifying the previous
one until a certain end stage is reached (here, the woman realizing the man’s malevolence). Finally,
a Partial transition relies on graphological and morphological patterns rather than layout and gram-
mar, since two domains are merged within a single panel rather than across a sequence, what previ-
ous studies term “blended-perspective” (Bimpikou 246; Maier and Bimpikou). Here, auxiliary
elements are added to the primary domain, such that we assume the narrative remains in the estab-
lished storyworld overall but overlays this scene with auxiliary aspects (the example here shows an
Imagined reflection within an otherwise primary domain environment).

Discussion
Altogether, the Domain Constructions (DC) model describes how graphic, combinatorial, and
conceptual structures contribute to construing incongruity in visual narrative sequences as depic-
tions of auxiliary domains, such as character’s (private) thoughts or other mental processes (e.g.
dreams). Inferences of such specific predicates thus signal that the incongruity was likely intentional,
meant to distinguish these events from (congruous) events belonging to the established, expected
storyworld accessible to all characters, i.e. the primary domain.
The DC model expands upon prior (semantic) analyses of intentional incongruity by arguing that
graphological, layout, morphological, and grammatical structures can all cue the conceptual inter-
pretation of an auxiliary domain. For graphic structure, these include correspondences between
meaning and colour schemes, (panel border) linework, or layout; within combinatorial structure,
morphological strategies distinguish domain switches or connect seemingly distinct figures, and
narrative patterns may align with domains. Interfaces across the three components then differentiate
how domains are presented in a sequence, characterized in three types of domain transitions. In
addition, these constructions appear together in visual sequences, inviting hypotheses related to the
(mis)alignment of varying cues, e.g. that the alignment of multiple cues could facilitate the inferencing
related to auxiliary domains.
As demonstrated in the range of examples throughout this paper, correspondences between form-
related cues and meaning appear to be encoded patterns across visual narratives, emphasizing how
incongruity forms a fundamental part of visual storytelling, in addition to congruity. Many visual
naratives are analyzed for their contingency across discourse units (Maier and Bimpikou; Abusch),
with discontinuity signalling event segmentation (Gernsbacher, “Two Decades” 5; Loschky et al.
322). However, the prevalence of incongruous information–semantic, graphic, and combinatorial–
supports that such constructions are also central within visual storytelling. In addition, this also
challenges the notion of a narrative as a temporal succession of events (Genette 25), as some depic-
tions of experiences in domain constructions do not align with the established timeline (e.g. depic-
tions of past/future events). While these can be construed as e.g. memories recalled in the current
timeline, such inferences follow resolution based on (temporally) incongruous depictions. The ques-
tion follows what effect such incongruity has on readers’ narrative experience. As these sequences
can still be comprehended well (Klomberg, Fadeeva, et al.) we speculate this complexity contributes
to a sense of intruiging, artful storytelling.
Furthermore, if indeed visual narrative comprehension relies on such patterned structures, in-
volving slots to be filled with appropriate content, it suggests authors and readers have learned these
patterns, and that these might be modulated by visual fluency, as shown for other aspects of visual
54 | JOURNAL OF COMPARATIVE LITERATURE AND AESTHETICS

languages (Cohn et al. 28; Cohn and Kutas 274; Cohn and Maher 79). Such learned constructions
may have changed over time or differ across genres or cultures. Some constructions may even
extend beyond comic narratives, depending on the affordances that these require. Colour differ-
ences, for instance, easily apply to film, while characteristics of panel borders remain a unique
graphologial feature of drawn images. Overall, these questions provide relevant challenges for fu-
ture corpus and/or experimental studies into the comprehension of domain constructions, within the
studies of comics, discourse, focalization, perspective taking, and cognitive linguistics.
All in all, the DC model provides a framework for constructions of domains in visual narratives,
where graphic, combinatorial, and meaning structures combine as formal parallel components under-
lying a holistic interpretation of incongruity within sequences. These structures capture the com-
plexity of visual narrative storytelling and its patterned nature, and may easily connect to how
domain constructions could operate across modalities. The implications of such patterns raise rel-
evant questions for future work into visual narrative comprehension.

Tilburg University, Netherlands

Acknowledgements

This project has received funding from the European Research Council (ERC) under the European Union’s
Horizon 2020 research and innovation programme (grant agreement No 850975).

Works Cited

Abusch, Dorit. “Applying Discourse Semantics and Pragmatics to Co-Reference in Picture Sequences.” Pro-
ceedings of Sinn Und Bedeutung, vol. 17, 2012.
Abusch, Dorit, and Mats Rooth. “Temporal and Intensional Pictorial Conation.” Proceedings of Sinn Und
Bedeutung, vol. 26, 2022.
Andrew, Clark, et al. Understanding Greek Vases: A Guide to Terms, Styles, and Techniques. Getty Publications,
2002.
Bimpikou, Sofia. “Perspective Blending in Graphic Media.” Proceedings of the ESSLLI 2018 Student Session,
ESSLLI, 2018, pp. 245–57.
Clausner, Timothy C., and William Croft. “Domains and Image Schemas.” Cognitive Linguistics, vol. 10, no.
1, 1999, pp. 1–31, https://doi.org/0936±5907/99/0010±0001.
Cohn, Neil. “Combinatorial Morphology in Visual Languages.” The Construction of Words, edited by Geert
Booij, vol. 4, Springer International Publishing, 2018, pp. 175–99, https://doi.org/10.1007/978-3-319-
74394-3_7.
—. “(Pea)Nuts and Bolts of Visual Narrative: Structure and Meaning in Sequential Image Comprehension.”
Cognitive Psychology, vol. 65, no. 1, Aug. 2012, pp. 1–38, https://doi.org/10.1016/j.cogpsych.2012.01.003.
—. The Visual Language of Comics: Introduction to the Structure and Cognition of Sequential Images. Bloomsbury
Academic, 2013.
—. “Visual Narrative Structure.” Cognitive Science, vol. 37, no. 3, 2013, pp. 413–52, https://doi.org/10.1111/
cogs.12016.
—. Who Understands Comics? Questioning the Universality of Visual Language Comprehension. Bloomsbury
Academic, 2020.
—. “Your Brain on Comics: A Cognitive Model of Visual Narrative Comprehension.” Topics in Cognitive
Science, vol. 12, no. 1, Jan. 2020, pp. 352–86, https://doi.org/10.1111/tops.12421.
 Constructing Domains in Visual Narratives | 55

Cohn, Neil, and Marta Kutas. “Getting a Cue before Getting a Clue: Event-Related Potentials to Inference in
Visual Narrative Comprehension.” Neuropsychologia, vol. 77, Oct. 2015, pp. 267–78, https://doi.org/10.1016/
j.neuropsychologia.2015.08.026.
Cohn, Neil, and Stephen Maher. “The Notion of the Motion: The Neurocognition of Motion Lines in Visual
Narratives.” Brain Research, 2015, p. 12.
Dijk, Teun A. Van, and Walter Kintsch. Strategies of Discourse Comprehension. Academic Press, 1983.
Duncan, Randy. “Persuasive Comics.” What Political Science Can Learn from the Humanities: Blurring Genres,
edited by R.A.W. Rhodes and Susan Hodgett, 2021, pp. 259–86.
Fauconnier, Gilles, and Mark Turner. The Way We Think: Conceptual Blending and the Mind’s Hidden
Complexities. Basic Books, 2002.
Forabosco, Giovannantonio. “Is the Concept of Incongruity Still a Useful Construct for the Advancement of
Humor Research?” Lodz Papers in Pragmatics, vol. 4, no. 1, Jan. 2008, pp. 45–62, https://doi.org/10.2478/
v10016-008-0003-5.
Genette, Gérard. Narrative Discourse: An Essay in Method. Translated by Jane E. Lewin, Cornell University
Press, 1980.
Gernsbacher, Morton Ann. “Coherence Cues Mapping during Comprehension.” Processing Interclausal Rela-
tionships in the Production and Comprehension of Text, edited by J. Costermans and M. Fayol, Erlbaum, 1997,
pp. 3–21.
—. “Two Decades of Structure Building.” Discourse Processes, vol. 23, no. 3, 1997, pp. 265–304, https://
doi.org/10.1080/01638539709544994.
Giora, Rachel, et al. “Weapons of Mass Distraction: Optimal Innovation and Pleasure Ratings.” Metaphor and
Symbol, vol. 19, no. 2, May 2004, pp. 115–41, https://doi.org/10.1207/s15327868ms1902_2.
Hacimusaoglu,
Hacımusaoğlu Irmak, et al. “Navigating Meaning in the Spatial Layouts of Comics: A Cross-Cultural Corpus
Analysis.” Visual Cognition, vol. 31, no. 2, Feb. 2023, pp. 126–37, https://doi.org/10.1080/
13506285.2023.2198271.
Horstkotte, Silke, and Nancy Pedri. Experiencing Visual Storyworlds. Ohio State University Press, 2022.
Huff, Markus, et al. “Cross-Codal Integration of Bridging-Event Information in Narrative Understanding.”
Memory & Cognition, vol. 48, no. 6, Aug. 2020, pp. 942–56, https://doi.org/10.3758/s13421-020-01039-z.
Hutson, John P., et al. “Understanding Moment-to-Moment Processing of Visual Narratives.” Cognitive
Science, vol. 42, no. 8, 2018, pp. 2999–3033, https://doi.org/10.1111/cogs.12699.
Jackendoff, Ray. Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford University Press, 2002.
Klomberg, Bien, Irmak Hacimusaoglu,
Hacımusaoğlu et al. “Continuity, Co-reference, and Inference in Visual Sequencing.”
Glossa: A Journal of General Linguistics, vol. 8, no. 1, 2023, pp. 1-43, https://doi.org/10.16995/glossa.998.
Klomberg, Bien, Klavdiia Fadeeva, et al. The Costs and Contours of Incongruous Identities in Visual Narratives.
2023. Department of Communication and Cognition, Tilburg University.
Langacker, Ronald. Foundations of Cognitive Grammar. Volume I: Theoretical Prerequisites. Stanford University
Press, 1987.
Loschky, Lester C., et al. “The Scene Perception & Event Comprehension Theory (SPECT) Applied to Visual
Narratives.” Topics in Cognitive Science, vol. 12, no. 1, 2020, pp. 311–51, https://doi.org/10.1111/tops.12455.
Magliano, Joseph P., et al. “The Relative Roles of Visuospatial and Linguistic Working Memory Systems in
Generating Inferences during Visual Narrative Comprehension.” Memory & Cognition, vol. 44, no. 2, Feb.
2016, pp. 207–19, https://doi.org/10.3758/s13421-015-0558-7.
Maier, Emar, and Soa Bimpikou. “Shifting Perspectives in Pictorial Narratives.” Proceedings of Sinn Und
Bedeutung, vol. 2, 2019, pp. 91–105.
Mikkonen, Kai. The Narratology of Comic Art. Routledge, 2017.
Ramachandran, V. S., and William Hirstein. “The Science of Art: A Neurological Theory of Aesthetic
Experience.” Journal of Consciousness Studies, vol. 6, no. 6–7, 1999, pp. 15–51.
Schilperoord, Joost. “Ways with Pictures: Visual Incongruities and Metaphor.” Visual Metaphor: Structure and
Process, edited by Gerard Steen, John Benjamins Publishing Company, 2018, pp. 11–47.
van der Gouw, Sharitha, et al. Visual Storytelling: A Corpus Analysis of Domain Usage in American and Japanese
Comic Styles. 2022. Tilburg University, Master’s thesis.
Willats, John. Art and Representation: New Principles in the Analysis of Pictures. Princeton University Press,
1997.
Zwaan, Rolf A., and Gabriel A. Radvansky. “Situation Models in Language Comprehension and Memory.”
Psychological Bulletin, vol. 123, no. 2, 1998, pp. 162–85.