Beyond the Metaphor: Conceptualizing Children’s Literature as (part of) a Rhizomatic Network

: If, as George E.P. Box puts it, »all models are wrong, but some are useful« (Box in Ahnert et al. 2020, 79), what then, would be the merit and concrete gains of such an ambivalent model in the field of literature? This article stems from a hunch: that the use of the network metaphor to describe children’s literature (in the broad sense as referring to any cultural product developed for children) is not insignificant. Starting from that postulate, the goal of this article is to look beyond the metaphor and explore how the rhizomatic network could serve as a concrete model, supplementing the current toolbox used to study children’s literature. Indeed, many characteristics of the rhizomatic network – namely its unlimited, simplified, non-hierarchical, random-access, and visual nature – lend themselves to a broader and more inclusive conceptualization of children’s literature. Translator study scholar Rebecca Walkowitz makes a strong case for this approach, stating that »[i]n the future, we will need to read comparatively, by which I mean reading across editions and formats and also recognizing that any one edition and format contributes to the work rather than exhausts it« (Walkowitz 2015, unpag.). Concretely, I argue for the use of the rhizomatic network as a visual model of multimodal children’s literature at three levels: 1) a given storyworld as a network of interconnected versions; 2) the context of any given version of the storyworld as a network; and 3) the text (or multimodal ensemble) of any given version of the story-world as a network of meaning-making resources (modes). I illustrate the network model at these three levels through two case studies: We’re Going on a Bear Hunt (Rosen/Oxenbury 1989) and the Gruffalo (Donaldson/Scheffler 1999). In Cathlena Martin’s


Introduction
In the fields of literature, translation, and adaptation studies, several researchers1 have used the term »network« to refer to the constellation of versions that constitute a storyworld,2 which seems an appropriate metaphor to map a rich and pluriform field.Indeed, as Deborah Cartmell points out, adaptation is »avowedly multidirectional and better represented by a network model: any node in the network may initiate an adaptation project in any direction« (Cartmell 2012, 133 sq.).Cecilia Alvstad and Alexandra Assis Rosa, as well as Outi Paloposki and Kaisa Koskinen make similar claims about the relationship between translations and retranslations.The former argue that »[a] network of affinities and differences may be mapped in the dialogic intertextual relations that any retranslation may establish with the sometimes rather complex web of texts that have been used as sources« (Alvstad/Assis Rosa 2015, 9 sq.), while the latter suggest that »the textual relations between different versions seem to form a ›rhizomatic‹ network of influences, ideologies and value judgments« (Koskinen/Paloposki 2015, 47).
The rhizomatic network (after Deleuze/ Guattari 1976Guattari , 1980;;Eco 2014) draws from the »subterranean plant stem that grows horizontally, sending out roots and shoots from its nodes« (Ahnert et al. 2020, 26).It is characterized by the multiplicity of entry and exit points (ibid.), and the absence of hierarchy between its constituents (Bermann/Porter 2014, 362 sq.).In addition, the rhizome does not have a centre, and lacks a periphery and polarised points (Katan 2018, 38).Instead, it is an unlimited system of connected nodes.Incidentally, my prior research already led me to analyse that same metaphor through an investigation of the labyrinth motif in Mark Z. Danielewski's experimental novel House of Leaves.I argued that the narrative and metanarrative labyrinths that Danielewski crafted could all be conceptualised as rhizomatic networks.Although my analysis was based on only one case study, I contend that the connection between literature and networks is neither accidental nor incidental.This article seeks to further explore this connection and, in doing so, highlight the potential of the rhizomatic network to study the relationships between literary texts and their translations and intermedial adapta-tions, their complex contexts of production and reception, and the ways in which they combine resources to make meaning.
Instead of focusing on literature at large, the present article limits its focus to multimodal children's literature.This particular object of study is the ideal site for a rhizomatic investigation, as it is particularly prone to various kinds of transformations, supported by »the increasingly common global circulation of mass media, and the ostensible collapse of markers of cultural hegemony, subcultures, and geo-temporal organization that comes with the accessibility and movement of text, performance, and image through social media, blogs, and webpages« (Geddes 2020, 424).In this article, children's literature refers to any product made for »the entertainment, exploitation, or enculturation of children« (Hunt 2001, 3).I will for instance consider films, video games, live performances, merchandise and YouTube videos.As products of children's literature.Regarding »multimodality«, I follow Gunther Kress and Theo Van Leeuwen's definition: »the use of several semiotic modes in the design of a semiotic product or event, together with the particular way in which these modes are combined« (Kress/Van Leeuwen 2001, 20).In that context, a mode is defined as »a socially shaped and culturally given resource for making meaning« (Kress 2009, 54).All works of literature are arguably multimodal (see Gambier 2006;Baldry/Thibault 2006;Gibbons 2012), however, children's literature in particular showcases a wide array of resources to make meaning (e. g., gestures, images, textures, and sound effects).
The rhizomatic network is an unlimited model that offers a simplified visual overview of a complex system and presents information as interconnected and non-hierarchical nodes that can be accessed randomly.In this article, I set out to explore what is to be gained by going beyond the metaphor and proposing a rhizomatic conceptualisation of multimodal children's literature.Concretely speaking, I argue for the use of the rhizomatic network as a visual model of multimodal children's literature at three levels: 1) a given storyworld as a network of interconnected versions; 2) the context of any given version of the storyworld as a network; and 3) the text of any given version of the storyworld as a network.Within the scope of this article, I use the terms »context« and »text« as defined by Elisabetta Adami and Sara Ramos Pinto in the context of multimodality.The former refers to »the social semiotic environment for the design, production, distribution/circulation of, and engagement with […] texts« and the latter to »any multimodally composed meaningful whole (or multimodal ensemble)« (Adami/Ramos Pinto 2020, 73).Although those three types of rhizomatic networks will be discussed separately, they are interconnected (see section 3).Indeed, as Umberto Eco argues, the outside of the network is still a network: »The rhizome has its own outside with which it makes another rhizome; therefore, a rhizomatic whole has neither outside nor inside« (Eco 1984, 82).This article is divided into three sections.Section 1 provides a series of preliminary comments on the limitations of using the rhizomatic network as a model to conceptualise multimodal children's literature.Section 2 introduces the broader context of the article through Robert Stam's concept of »intertextual dialogism« (2000).Section 3 discusses the features of the rhizomatic network and their concrete applications to the three different kinds of rhizomatic networks outlined in the introduction (storyworld, context, and text).In all three sections, I provide examples and illustrations by means of two children's literature case studies: We're Going on a Bear Hunt (Rosen 1989) and the Gruffalo (Donaldson 1999).»We're Going on a Bear Hunt« is an American folk song that has been sung in scout camps since the 1970s.It recounts the adventure of a group across a series of natural obstacles, in search for a bear.Once they come across the bear, the group runs backwards through the obstacles.The song was adapted into a picturebook by Michael Rosen, featuring watercolour illustrations by Helen Oxenbury that represent a group of children, a baby, and their dog.The Gruffalo, in its best-known form, is a picturebook written by British author Julia Donaldson and illustrated by Axel Scheffler.The story revolves around a mouse and its successive encounters with a fox, an owl, and a snake.Each time, the predator wants to lure the mouse to eat it, and each time, the mouse scares it away by referring to its meeting with an imaginary monstruous creature -the Gruffalo.Halfway through the story, however, the mouse comes across a real Gruffalo.This time, the mouse pretends to be the scariest creature of the wood.As the Gruffalo follows the mouse and meets with the three predators again, they run away afraid, not of the little mouse but of the giant monster walking behind it, effectively tricking the Gruffalo into believing that the mouse is indeed the scariest creature of the woods.

Limitations and Positioning Bias
In this article, I propose a series of figures illustrating the network model.I do so, however, with the awareness that network visualisations are ambivalent: they »are often critiqued as either overly reductive (in terms of content and context) or absurdly illegible (in terms of visual complexity), yet they bring an immediacy to our perception of information« (Ahnert et al. 2020, 68).I therefore aim to strike a balance between providing useful information while offering a legible and simplified visual model.
Simplification, in this case, also means resorting to the default mode of academic expression -text -while at the same time underlining and reaffirming the importance of other modes and media in the creation of meaning.3In this article, I opted for nodes (containing words) arranged rhizomatically (connected through links), which is arguably already a multimodal manner to present information.However, it could be more so, by including the actual product (or a portion thereof) in the nodes (e. g., screenshots, audio files, videos, pictures).In Figure 1,4 I propose a sample of what such a rhizomatic network would look like using the Gruffalo franchise.Note that the textual description of the nodes could be removed but have been kept here for the sake of argument and clarity.The use of arrows is further explained in section 3.In spite of its apparent simplicity, the rhizomatic network is neither neutral nor straightforward.Just like taxonomies it does not fully escape »the prejudices and partialities of their inventor« (Cardwell/Whelehan 2010, 21).Though decentred in theory, the network is constructed and studied by a researcher situated in a geographical, temporal, cultural, linguistic, social, and economic context.It stems from a series of choices (conscious or not) that will inevitably have an influence on the way in which we read and interpret the network (Ahnert et al. 2020, 63).As Linda Hutcheon points out: The decentering of our categories of thought always relies on the centers it contests for its very definition […] the spreading rhizome might be a less repressively structuring concept than the hierarchical tree.But the power of these new expressions is always paradoxically derived from that which they challenge.(Hutcheon 2008, 60) While the definition of children's literature mentioned in the introduction recognizes that it encompasses more than printed books, my background in literary studies means that the literary artefact is the starting point of my research, which has repercussions on the networks that I create and how I frame information.In the Bear Hunt storyworld network in section 3, the picturebook constitutes is fairly central hub within the network insofar as my knowledge of its intermedial adaptations and translations is more extensive that those of -for instance -the film or the song.Such a conundrum is hardly avoidable, though increased and sustained interdisciplinary collaborations can help diversify perspectives on the model.The Bear Hunt network that I propose in this article is not fixed, and will evolve as new versions of the work appear, new knowledge is found, and new collaborations arise.In line with Pierre Lévy's notion of »collective intelligence« (Jenkins 2006, 4), i. e. »the ability of virtual communities to leverage the knowledge and expertise of their members, often through large-scale collaboration and deliberation« (ibid., 281), the network model encourages interdisciplinary discussions in order to develop its full potential.

Intertextual Dialogism
In her article »Charlotte's Website: Media Transformation and the Intertextual Web of Children's Culture«, Cathlena Martin explores the different versions of E.B. White's Charlotte's Web: For many years, the original 1952 novel by White comprised the entirety of the story of Charlotte's Web.Today, however, stories do not stay confined to one medium.The Hanna-Barbera animated film was released in 1973, with a sequel in 2003: Charlotte's Web 2: Wilbur's Great Adventure.Finally, in 2006 the live action film was released, along with a video game.These print and media versions converge to create a larger ur-text.(Martin 2009, 87) While Martin convincingly makes a case for a rhizomatic conceptualisation of children's literature, she still refers to the 1952 version of the story as the original, a notion also fostered by the use of »ur-text«.I argue that placing the work in Robert Stam's broader context of »intertextual dialogism«, i. e. the »infinite and open-ended possibilities generated by all the discursive practices of a culture, the entire matrix of communicative utterances within which the text is situated« (Stam 2000, 64) draws attention to the multidirectionality of the network.In the case of Charlotte's Web it includes sources of influence such as Willis J. Gertsch's American Spiders (1949) and John Henry Comstock's The Spider Book (1948) (Neumeyer 1995, 225), as well as prior unpublished versions that shed new light on the development of the storyworld.For instance, chapter 3 of the published version of Charlotte's Web was initially supposed to be its introduction and the first draft did not feature the human characters as prominently (Elledge 1984, 295).
The entire matrix of communicative utterances also draws attention to the afterlife of the work.In Figure 2, Jennifer Miskec develops the network of the Fancy Nancy franchise, which includes reviews, scholarship on the text, and scholarship on the text's transformations.This highlights the notion of prosumers, i. e. »individuals who consume and produce value, either for self-consumption or consumption by others, and can receive implicit or explicit incentives from organizations involved in the exchange« (Lang et al. 2020, 178), and what Jenkins calls »participatory culture«, a culture in which: »fans and other consumers are invited to actively participate in the creation and circulation of new content« (Jenkins 2006, 290).Example of consumer-produced versions of products of children's literature include lesson plans, recipes, fanart, school activities, and arts and crafts projects.The present article is, therefore, also part of the network of the Gruffalo and that of We're Going on a Bear Hunt.All in all, intertextual dialogism provides the broader context in which the three kinds of networks (storyworld, context, and text) are situated, which aligns with »an inclusivist conception of adaptation as a freewheeling cultural process: flagrantly transgressing cultural and media hierarchies, wilfully cross-cultural, and more weblike than straightforwardly linear in its creative dynamic« (Murray 2012, 2).
While infinite and open-ended in theory, in practice, only a portion of the network will be examined, according to the scope and focus of the research (e. g. translations, intramedial adaptations, merchandise, contemporary rewritings).

Storyworld
In spite of »embrac[ing] intertextual equality, reject[ing] old hierarchical distinctions, and seek [ing] to blur the boundaries between adaptations and related forms« (Cardwell 2018, 12-3), much of the adaptation studies scholarship has focused on novel-to-film adaptations, and children's literature scholars are no exception (McCallum 2018;Hermansson 2019;Meeusen 2020).Commenting on Hutcheon's A Theory of Adaptation (2006), Kamilla Elliott suggests that »opening the field beyond the usual suspects of literature, theatre and film to include ›video games, theme park rides, Web sites, graphic novels, song covers, operas, musicals, ballets, and radio and stage plays‹« results in fluid and »unclear borders and boundaries« (Elliott 2013, 30-1).Elliott concludes that »Hutcheon's theory is more pastiche than system« (ibid., 31), copy-pasting method from one field to another instead of accounting for the specifics of the adaptation at hand.Such an undertaking, though desirable, is not without risks, as Deborah Cartmell and Imelda Whelehan warn »in the attempt to anticipate every possible permutation of the relationship between one storyworld form and another we attempt a list that will never be exhaustive but is, frankly, exhausting« (Cartmell/Whelehan 2010, 22).
The rhizomatic network has the merit to propose an overview of as many versions of a storyworld as one can collect and connect, not in an attempt at exhaustivity, but rather awareness and inclusivity of even the more borderline cases.For instance, Hutcheon wonders: »would a museum exhibit be an adaptation?« (Hutcheon 2008, 172).From the perspective of the rhizomatic network, a museum exhibit would be a node like any other, connected to the work(s) from which it derives, yet situated in its own unique context of production and reception.Figure 3 gives an inexhaustive overview of different versions of the Bear Hunt storyworld that I have been able to collect.Though the rhizomatic network is a random-access and non-hierarchical model, indicating chronology by means of arrows is not necessarily incompatible with those features.While the arrows provide more context on the production processes and signal directions of influence (after Blankier 2014, 110), the random-access nature of the rhizome emphasises that it has no bearing on the audience's consumption processes.Instead of making an attempt at terminological clarity and uniformity, the network foregrounds the idea of connection and frames intermedial adaptations and translations as »processes in which connections are established between two different modes of representation« (Schober 2013, 89), thus creating a »new set of relations« whereby »something unique is produced« (Cartmell/Whelehan 2010, 22).In that respect, transmedia storytelling is a particularly relevant concept when addressing the way(s) in which the new product is unique, as it refers to »stories that unfold across multiple media platforms, with each medium making distinctive contributions to our understanding of the world, a more integrated approach to franchise and development than models based on urtexts and ancillary products« (Jenkins 2006, 293).It should nonetheless be mentioned that »transmedia« does not fully account for all the versions of the work in the rhizome.The French dub of an animated film, for instance, does not change medium.In that case, the storyworld does not cross medial but linguistic and cultural boundaries.In that respect, Klaus Kaindl's translation taxonomy (2013) offers a broader overview of modal and linguistic shifts.Kaindl uses the term »translation« as encompassing adaptation processes, and his model consists of intramodal intracultural translation, intramodal transcultural translation, intermodal intracultural translation, and intermodal transcultural translation.This model could be supplemented with the parameters »intramedial« and »intermedial« in order to incorporate the change of medium that is suggested by the concept of transmedia storytelling.According to that model, the French dub of the Gruffalo animated film would be an intramodal (within the same mode; aural), transcultural (in a different language; from English to French), intramedial (within the same medium; animated film) translation of the English dub.
Not only is each version of the storyworld different in terms of parameters such as medium, language and genre, but there is also »a wider communicative context [to adaptations] that any theory of adaptation would do well to consider« (Hutcheon 2006, 26), which includes temporal, geographical, ideological, and socio-cultural factors.The complexity of that context cannot be understated.As Colin MacCabe points out: »the number of variables involved in any adaptation approach[es] infinity« (MacCabe 2011, 8).
From the consumer's perspective, the experience of the network is much different.For one, the awareness and construction of an »adaptation« arises from the consumer's knowledge of the network of a particular storyworld.Regina Schober argues: »In order for an adaptation to be identified as such, and to acquire its status as an adaptation, there must be an active process of forming specific connections to a reference medium either on the production or the reception side« (Schober 2013, 90;my emphasis).This point can also be made regarding translations: although paratextual information (usually) provides somewhat clear indications that a picturebook is a translation, not all readers -especially children, even more so if they cannot yet read -will be aware of it.Media such as films and video games require knowledge of their production context or a thorough reading of the DVD box or video game credits to be aware of their status as translations.
The network exists as a passive construct, but becomes an active and organic entity as the consumer makes their way through the different versions of a work, effectively making connections.And even then, its random-access quality and the fact that »[e]ach … entry needs to be self-contained« (Jenkins 2006, 96) mean that the consumer can access any number of nodes in any order, including only accessing one.However, »[r]eading across the media sustains a depth of experience that motivates consumption« (ibid.).This goes back to Martin's example: any version of the Charlotte's Web is Charlotte's Web, and the combinations of all versions of Charlotte's Web is Charlotte's Web.While forces such as canonisation, fidelity and nostalgia still hold sway on the part of the audience and tend to place some versions of the storyworld above others in terms of prestige and/or authority, the rhizomatic network is non-hierarchical.As such, it reflects the experience of »consumer children who no longer view the printed text as the only way to experience [the work, because] multimedia adaptations have been the normal publishing practice in their lifetime« (Martin 2009, 88), thus framing each version of the storyworld less as »a resuscitation of an originary word than a turn in an ongoing dialogical process« (Stam 2004, 25).

Context
According to Adami and Ramos Pinto, the context of a work includes »participants in each of these processes [design, production, distribution/circulation], and the social and semiotic resources available to make signs and meanings« (Adami/ Ramos Pinto 2020, 73).In Figure 4, I propose a visual representation of the context of the Bear Hunt film, which includes individual participants such as Michael Rosen (as scriptwriter), Helen Oxenbury (as scriptwriter), Joanna Harrison (as filmmaker and scriptwriter), Robin Shaw (as filmmaker), and Olivia Colman (as voice actor), as well as organisations as participants: Channel 4 (the network broadcasting it) and Lupus film (the studio producing it).Adami and Ramos Pinto further argue that social and semiotic resources are tied to the medium and its context of production and distribution (in this case: animated film broadcast on TV).To further illustrate that networks are not self-contained but rather interconnected with other networks, Michael Rosen and Helen Oxenbury are also part of the context of the picturebook, respectively as author and illustrator.Both are also part of the context of the song, Rosen through his live shows and Helen Oxenbury as the illustrator of Alison McMorland's vinyl cover.In addition, Joanna Harrison started her career as a storyboard artist on Raymond Brigg's adaptation of The Snowman (1978), which is connected to the Bear Hunt through her, as well as through their genre (Christmas film) and broadcasting context (films broadcast on Christmas Eve on Channel 4).At the level of the text, the use of a scarf as a token of remembrance tied to loss and grief in the Bear Hunt film may be an intertextual reference to a similar use of the scarf in the Snowman film.Those common traits result in the two films often being compared or mentioned alongside one another.This brief example illustrates Karen Coats's statement that »the site of a text's production, form, and audience response are webbed and mutually informing« (Coats 2017, 222).
As mentioned in section 1, network representations are simplified and fail to provide certain information.Notably, crucial spatio-temporal information is missing.Such information is important in order to gain a better understanding not only of the full context of a given node, but also of the relations between nodes.For example, whether a retranslation is »hot« or »cold«;5 whether technological advances need to be taken into account between the film and the video game; whether the initial audience of the picturebook is the same as that of the film.
Conceptualising the context of production and reception of children's literature as a network also helps shed light on the notion of »core« of a story (sometimes also expressed as »essence« or »spirit«).While such elusive terms suggest that the core is a quality inherent to the storyworld, a study conducted by Barbara Herrnstein Smith in 1980 on 345 variants of Cinderella led to the conclusion that: »it was difficult to establish a common core beyond a very general level.[…] even if there was some consensus […] this would be because the people producing this basic story had a similar background, and similar purpose« (Brownlie 2006, 149;my emphasis).This displaces the idea of core from the storyworld and its different versions to their respective contexts of production and reception.Much like the hole in a donut is a hole because of what is around it, what creates the similar core of a storyworld is a similar context, not an inherent quality.

Text
Moving from the context to the text -or multimodal ensemble -Hartmut Stöckl makes the argument that multimodality can be modelled as a »networked system of core modes, medial variants, peripheral modes, sub-modes« (Stöckl 2004, 9).However, his proposed model is »a hierarchically structured and networked system, in which any one mode can be seen to fall into sub-modes which in their turn consist of distinct features that make up the sub-mode« (ibid., 12; my emphasis).More recent models, such as Pérez-González's (2014), follow a similar hierarchical structure.It is not so surprising that hierarchy is inherent to multimodality, as multimodality func- tions with broader categories and more fine-grained ones in order to provide the necessary nuance and precision.For instance, the aural category encompasses the modes »voice«, »music«, and »sound effects«, and the mode »voice« can be divided into sub-modes including »pitch«, »volume«, »tone«.Across levels, however, no node is more important than the others.It is especially important because, just like the printed text, the textual mode has benefited from more attention and prestige, as it supposedly has »the highest ›reach‹«, the ability to »serve the widest range of communicative functions« and to enable »the highest, most complex forms of thinking« (Jewitt/Bezemer/O'Halloran 2016, 3).This assumption underpins part of the book-to-film discourse in adaptation studies, which focuses on what media can and cannot do (see Lindsay 1915;Bluestone 1957;Chatman 1980).However, researchers working with multimodality have pointed out that it is not so much about which mode is »better« than the others, but about the different meaning-making potentials offered by each (Jewitt/Bezemer/O'Halloran 2016, 3).By being non-hierarchical, the rhizome model visually emphasises that modes exist alongside one another and in combination with one another, and therefore resists text-centred approaches.While, in practice, multimodal ensembles are experienced as »wholes«, it is useful for researchers in the field of multimodality to be able to have an overview of the different resources involved in the creation of a given semiotic product.As such, the random-access nature of the rhizomatic network makes it possible to focus on any element of that whole, or combinations thereof, depending on the research at hand.
Since one of the goals of this article is to propose a useful way of conceptualizing multimodality as a rhizomatic network, it is important to acknowledge the other forms that naturally emerge and that could supplement that representation.In Forms, Caroline Levine evokes four different forms that are »particularly common, pervasive -and also significant […] political structures that have most concerned literary and cultural studies scholars« (Levine 2015, 21): whole, rhythm, hierarchy, and network.Conceptualising multimodality as a hierarchy and as a whole offer complementary approaches, insofar as the former reflects the top-down, theoretical breakdown of meaning-making resources into categories and sub-categories, while the latter reflects the experience of the consumers, who experience the »synergy« (Sipe 2012) of modes and the meaning resulting from their combinations.
As mentioned previously, several typologies of modes and sub-modes exist depending on their field of application.In this article, I makes use of the following modal categories: visual, textual, aural (Oittinen/Ketola/Garavini 2017), as well as tactile and olfactory.In order to refine my analysis, I also use the term »sub-modes« (Stöckl 2004;Pérez-González 2014).For instance, an image is a visual mode, and its sub-modes include colours, lines, shading and perspective.
Since meaning arises from the synergy of all modes and sub-modes, the connections between them are a given.It is, therefore, not useful to represent those in practice.Instead, I propose to use the rhizomatic network model to highlight salient connections between (sub-)modes in order to foreground interesting dynamics and combinations.The rhizomatic network makes it possible to highlight certain connections, combining bottom-up and top-down approaches.Through the former, I observe the resources used in a specific product of children's literature and how they connect to create meaning.Through the latter, I consider the rhizomatic network of modes through a specific lens, or research question.Looking at the scene in which the mouse meets the Gruffalo through the network model illustrated below (Figure 5 and 6), I want to demonstrate that the multimodal arrangement of the scene strengthens the threat that the Gruffalo represents.
The hybrid model based on Bar-Yam's (2002) combines hierarchical features with lateral rhizomatic connections.Figure 5 shows the modal categories and the sub-modes, and highlights lateral connections.Figure 6 removes the hierarchical aspect.As a result, some nodes appear to be floating or to not be connected to all the nodes.This is because of the postulate underpinning this model, i. e. that all modes are connected to begin with (bottom-up).I have therefore only mapped the connections relevant to my study (top-down).
The three nodes at the top represent the main modal categories combined in that scene.To make meaning, it uses textual, aural, and visual resources, in this case: text, voice and illustrations.Their respective sub-modes are connected to them through dotted arrows while lateral connections between sub-modes are represented by full arrows.The arrows, once again, highlight the direction of influence.For instance, a sudden change in font size can influence the volume of the reader's voice.Likewise, the rhythm encoded in the text will prompt a certain reading rhythm.The character design, their facial expression, body language, attributes, and physical appearance can prompt certain intonations or voices: the Gruffalo is likely to speak in a more booming voice than the little brown mouse.In this part, I would like to invite the reader to dig their copy of the Gruffalo from the dusty box in their basement or attic and follow along.Looking at the composition, the Gruffalo is in the foreground, in immediate proximity to the mouseand to the reader.The creature looks scary (character design, reinforced by a bird flying away from the Gruffalo) and hungry (facial expression; its tongue is out).The mouse pauses (the shadow under the mouse almost suggests that it stopped midstride, or alternatively that it jumped in fear) and looks afraid (body language and facial expression) as the Gruffalo walks in its direction (composition), and against the reading direction, thus blocking both the mouse and the reader in their idle progression.The layout of the text is uneven: there is a lot more text on the left (mouse's side) than on the right (Gruffalo's side), which conveys a sense of urgency.The reader has time to read the lengthy text under the mouse but once they reach the side of the page where the Gruffalo is, there is very little text, as the threat is imminent.The length of the text, combined with the imminence of the threat may also influence the reading speed of the text (reading faster as the Gruffalo approaches).The metric used in the written text prompts readers to read rhythmically, while the sense of urgency conveyed by the mouse's body language can increase reading speed.The change of font size on the »Oh!« (and punctuation) suggests surprise.In addition, the character design and facial expression of the Gruffalo might prompt metacommentary during the read-aloud process, reinforcing its threatening presence, and so might the mouse's body language and facial expression.
While a bottom-up approach emphasises the wealth of modes and sub-modes involved in the meaning-making process, a top-down approach helps refine it and focus on the nodes and connections that are relevant for the question at hand.In this case, I have decided not to include such sub-modes as perspective, lines or colours, as they were less relevant to my analysis.

Conclusion
This article started from a hunch, that the network metaphor often used to refer to children's literature could offer a concrete model to conceptualise it.Within the scope of this article, I aimed to propose such a conceptualisation at three different, yet interrelated levels: that of the expanded storyworld, that of the context of each version, and that of each version itself as a multimodal ensemble.While those three levels can be viewed separately, I have shown that there is merit in considering their relations.The expansion of the storyworld across media and languages gives rise not only to new versions but also -for each of them -to a new context of production and reception.Notably, this includes a new relationship to the audience and the necessity for producers to negotiate making new with the old, confronting such forces as canonization, nostalgia, or the audience's pre-existing knowledge and resulting expectations in the process.In addition, multimodality provides a framework to systematically study modes and sub-modes and observe how they combine to make meaning.A multimodal approach that is sensitive to the context of production and reception of the work, as well as of its linguistic, medial, and generic conventions helps understand what happens to the multimodal ensemble when a given storyworld undergoes transformations.Placing all three of these networks within the broader context of Stam's intertextual dialogism highlights the virtually endless wealth of material and connections, and sheds light on transformations of the storyworld that have -thus far -benefited from less attention.It also blurs the line between producers and consumers and, by emphasising the open and inclusive nature of children's literature, foregrounds the latter as prosumers.A limitation associated to such a broad framework, however, is that the network does not contribute to terminological clarity in the fields of adaptation and translation studies.Instead, in a similar effort to that of Cartmell and Whelehan, it seeks to offer a greater awareness of the new sets of relations that ultimately produce something uniquethe same, but different.It observes the work from the inside (connections between modes), from the outside (connections between contextual elements) and in relations to other works (connection between versions of the storyworld) and thereby seeks to do justice to its complexity.In spite of the narrow focus of this article, I argue that the network model can be used to discuss products of literature at large, a field in which it might be even more pressing to consider all transformations undergone by the text and recognize that -although the printed medium and the textual mode are prevalent -novels make meaning in a number of other ways.Although inevitably imperfect the rhizome model promises exciting new ways to look at children's literature -and literature at large -in all its complexity.

Fig. 3 :
Fig. 3: The network of some Bear Hunt versions.

Fig. 4 :
Fig. 4: The context of the Bear Hunt film as a rhizomatic network.

Fig. 5 :
Fig. 5: Modal and sub-modal combinations in a scene from the Gruffalo.