Transmedia characters: Theory and analysis

This article sketches a theoretical framework and method for the analysis of transmedia characters that focuses on specific instantiations of these characters in individual media texts, before asking how these local work-specific characters relate to other local work-specific characters or coalesce into glocal transmedia characters as part of global transmedia character networks, thus evading what one could consider an undue emphasis on the “model of the single character” when analyzing the various characters that are, for example called Sherlock Holmes, Batman, or Lara Croft. The connections between these work-specific characters within transmedia character network could then be described as either relations of redundancy, relations of expansion, or relations ofmodification – with only redundancy and expansion allowing for medial representations of work-specific characters to contribute to the representation of a single transmedia character. In intersubjectively constructing characters across media, however, recipients will not only take into account powerful normative discourses that police the representation of characters across media but also draw on their accumulated knowledge about previously represented work-specific or transmedia characters as well as about transmedia character templates and even more general transmedia character types.

tural icons" (see Brooker 2013) that exist "above" any particular medial representation or media text. As will become clear, however, this is certainly not the only way to think about characters across mediaand the present article aims to complement rather than contradict approaches that emphasize other aspects of this rather complex phenomenon.
The first section, "Conceptualizing characters," discusses a necessarily narrow selection of the existing research on characters and sketches some of the different ways in which scholars have theorized what a character is or can be. The second section, "Constructing characters," zooms in on some of the medium-specific as well as transmedial strategies of representation and processes of comprehension that form the foundation on which we talk about characters in the first place. The third section, "Correlating characters," explores how characters that are represented in different media texts can relate to each other or, indeed, be understood as a single character under certain conditions, briefly hinting at some of the ways in which the construction and correlation of characters are subject to powerful normative discourses that involve both producers and recipients.
Certainly, there would be much more to say on every one of these questions and the other articles collected in this special issue contribute their own answers, often differing quite substantially in theoretical-conceptual and methodologicalepistemological orientation. One way or another, though, I hope that this brief explorative article still succeeds in presenting a perspective on the theory and analysis of characters across media with at least some heuristic value.

Conceptualizing characters
Despite the aforementioned saliency of characters in contemporary media culture, Jens Eder's early observation "that many have written only a little and only a few have written much on characters" (2008a: 40; "dass viele nur wenig und nur wenige viel zum Bereich der Figur geschrieben haben") still rings true today. The situation is further complicated by the fact that characters can not only move across media borders but also be more generally understood to be transmedial phenomena (see, e. g., Rajewsky 2002;Ryan 2006; Thon 2016 on the transmediality of phenomena such as characters). On the one hand, then, research that is not primarily concerned with characters may very well offer valuable contributions to their theory and analysis. On the other hand, and no less importantly, characters can be realized in a range of conventionally distinct media and they are realized differently in these media, necessitating not only general transmedial but also medium-specific theoretical perspectives (see, e. g., the contributions in Eder et al. 2010b; Leschke and Heidbrink 2010; Riis and Taylor 2019).

Transmedia characters: Theory and analysis
Beyond the substantial body of research on the representation of characters in literary texts (see, e. g., Jannidis 2004;Phelan 1989;Schneider 2000) and films (see, e. g., Eder 2008a;Smith 1995;Tomasi 1988), recent years have also increasingly seen the emergence of theoretical work on characters in other media, from comics (see, e. g., Aldama 2010; Varis 2019) via television series (see, e. g., Mittell 2015: 118-163;Pearson and Davies 2014: 149-184) to video games (see, e. g., Schröter and Thon 2014;Vella 2015). While there are differences between the ways in which literary texts, comics, films, television series, video games, and other conventionally distinct media forms represent characters, at first glance there seems to be a broad consensus that characters cannot be reduced to "textual effects" or "actantial functions" (see, e. g., Greimas 1983;Propp 1968;Tomasi 1988), but should be understood as "textor media-based figure [s] in a storyworld, usually human or humanlike" (Jannidis 2014: 30).
This comparatively simple definition seems to work well enough in many cases, but it also raises two questions, namely: What are the conditions under which we are prepared to describe a character as being located in a storyworld? And: What are the conditions under which we are prepared to describe a character as "human or human-like"? Perhaps, such a definition is too simple then, at least for our current purposes. 1 But, of course, there are other options: According to Eder, characters "are set apart from the other elements of fictional worldsrefrigerators, mountains, treesby their intentional (object-related) inner life; that is, by having perceptions, thoughts, motives, and emotions" (2010: 17). Against that background, Eder proposes to "envisage film characters as identifiable fictional beings with an inner life that exist as communicatively constructed artifacts" (2010: 18, original emphases). While I remain unconvinced that characters necessarily have to be fictional and thus would consider it less problematic to conceptualize them as represented rather than fictional entities within represented rather than fictional worlds, 2 I would agree that characters are more appropriately defined via their "intentionality" than via their "human-ness." 1 To be fair, this is meant to be a simple definition that opens a survey article in the influential Handbook of Narratology and I certainly do not mean to accuse Jannidis of being simplistic, not least because he explicitly qualifies characters' supposed anthropomorphicity ("usually"). See also the more nuanced discussion in Jannidis 2004;Eder, Jannidis, and Schneider 2010a. 2 While this is somewhat removed from the concern of the present article, I would still maintain that allowing for the possibility of nonfictional characters enables us to more clearly distinguish between the intersubjective communicative constructs of human or human-like entities that are represented nonfictionally and the actual entities these intersubjective communicative constructs are meant to be similar to. For further discussion of why we should not conflate what nonfictional representations represent with the "actual world," see, e. g., Thon 2014, Thon 2019 Which leaves us with the question to what extent our conceptualization of characters should require them to be entities located in a represented world or storyworld 3 of some sort. It seems safe to say that this is, either explicitly or implicitly, a very common part of current conceptualizations of characters, particularly in the context of the theory of fiction and/or narratological approaches to character analysis (see, e. g., Jannidis 2004;Eder 2008aEder , 2008bPhelan 1989). At the same time, however, this question leads us into the core conceptual and terminological quagmire of much current research on characters across media, namely that the term "character" as well as names such as "Sherlock Holmes," "Batman," or "Lara Croft" are used to refer to two very different kinds of phenomena, only one of which could be appropriately described as represented entities with an intentional inner life that are located in storyworlds. Tony Bennett, for example, frames James Bond not as a character but as a "popular hero" (2017: 1) and a "hero figure" (2017: 2) that is disconnected from any particular medial representation and has "assumed a semi-autonomous and quasi-real character, functioning as a 'free floating' signifier" (2017: 8; see also Bennett and Woolacott 1987;and the contributions in Lindner 2003). Similarly, Will Brooker analyzes Batman as a "cultural icon" (2013: 8) that is best understood as an "inherently multiple [...] amalgam of all his [Batman's] different forms, stories and histories" (2012: 151). Brooker likens this "amalgam" to a "myth that comprises all his contradictory variants, but is loose and flexible enough for the contradictions to not matter" (2012: 153; see also Brooker 2013; as well as the contributions in Pearson and Uricchio 1991;. 4 While characters' names (such as "James Bond" or "Batman") are commonly used to refer to "popular heroes" or "cultural icons," then, it is worth noting that neither Bennett nor Brooker explicitly and consistently conceptualize these complex cultural constructs as characters per se.
3 Taking into account that the kinds of medial representations that represent worlds can generally be considered to be narrative representations (see, e. g., Herman 2009: 9-22;Ryan 2006: 6-12;Thon 2016: 26-30), I do not see much of a difference between "represented worlds" and "storyworlds," but I would prefer either term to "fictional worlds" or "imaginary worlds" in order to stress that characters can be represented by nonfictional medial representations (see above). 4 Brooker does not leave it at that, but distinguishes the "myth" from the "brand" and the "canon" of Batman. The "brand" of Batman that refers to "a smaller, more contained and more controlled network of texts, defined by their current status as Warner Bros. Batman products: expressions of the contemporary template, rather than a broader, folk identity" (2012: 153); the "canon" of Batman refers to "the rulebook of continuity" and "the strict sense of what counts and what happened, what is 'true' and what isn't, in the mainstream Batman comic book universe" (2012: 154). While I cannot explore these complex and historically situated issues in any detail, I will briefly return to the kinds of normative discourses that are connected to these complementary frameworks of the "myth," the "brand," and the "canon" in the final section.

Transmedia characters: Theory and analysis
At first glance, this seems to be different in recent works at the intersection of Japanese studies and media studies on the function of characters in the Japanese "media mix," but even those works echo the broad distinction between characters as represented entities with an intentional inner life that are located in storyworlds, on the one hand, and characters as "popular heroes" or "cultural icons," on the other. Mark Steinberg, for example, is certainly less interested in issues of medial representation than he is in the historically situated production and consumption of "characters," which also leads him to ignore some of the intricacies of characters' medial representation discussed in the following section, but he still stresses "[t]he double nature of the character [that] allows it to function as the glue between divergent series: it is both a series of material embodiments and the immaterial entity that traverses and binds them. [...] Hence the character is both lodged within a particular material incarnation and constantly in excess of it" (2012: 195).
As Lukas Wilde (2018aWilde ( , 2018bWilde ( , 2019aWilde ( , 2019b has recently reconstructed in significantly greater detail, however, the Japanese discourse in fact offers a sophisticated terminological and conceptual apparatus to think through some of the issues connected to the representation of characters across media. Drawing on influential Japanese cultural theorists such as Azuma Hiroki (2001Hiroki ( , 2007, Itō Gō (2005; see also Itō et al. 2007), and Odagiri Hiroshi (2010), Wilde distinguishes between the kyarakutā (character) as "a fictitious being represented to exist within a diegetic domain (storyworld)" (2019a: 4-5) and the kyara as "a stylized or simplified visual figuration that can be easily reproduced and consumed outside of its original narrative context" (2019a: 5). While it seems clear from Wilde's careful reconstruction that the concept of kyara has much to offer to our understanding of characters across media, 5 the very distinction between kyara and kyarakutā would suggest that the former are not to be conflated with characters on either a conceptual or a terminological level. 6 5 Perhaps most saliently, Wilde notes "that kyara function very much like fictitious actors, playacting or performing a number of incoherent fictional roles" (2019a: 6), which provides an interesting alternative model to think about "popular heroes" or "cultural icons" and also serves as a welcome reminder that even locally represented characters can of course not only be able to achieve the "physically" and "humanly impossible" (Alber 2016: 25) but that their medial representations may also play with logical or representational impossibilities in various ways. 6 While I highly value Wilde's theoretical work, I am not completely convinced by the way he seems to conflate kyara with "pre-narrative" and "meta-narrative characters," nor would I equal "a decontextualized, trans-fictional, trans-world, orwith Azumameta-narrative entity" with "the sum of all Batmen (or Nick Furys)" (Wilde 2019a: 10). While "popular heroes," "cultural icons," or indeed "meta-narrative characters" are not the main focus of the present article, it would seem to Yet, even those (few) theorists that explicitly and consistently use the term "character" to refer to the complex cultural constructs that I have attempted to circumscribe throughout the previous pages (and that go significantly beyond represented entities with an intentional inner life that are located in storyworlds) tend to acknowledge that there is a distinction to be drawn here. Perhaps the most salient example of this can be found in Paolo Bertetti's work on transmedia characters, as he distinguishes between "characters based on a single course of events" and "characters based on multiple courses of events" (2014: 2350; see also Bertetti 2019). Again, the main query I would have here is why we would want to use the term "character" to refer to these very different phenomenaand this seems even more puzzling to me since other terminological options are readily available. Shane Denson and Ruth Mayer, for example, draw a similar distinction between "series characters [...] in the more or less closed fictional universe of a serially-ongoing narrative" (2018: 67) and "serial figures" that "are shaped and reshaped through the repetitions, revisions, and reboots of their stories" (2018: 68; see also Denson 2014;Denson and Mayer 2012;Mayer 2016).
While I would readily acknowledge that Bertetti has made important contributions to our understanding of characters across media, it would seem to me that Denson and Mayer's conceptualization of "serial figures" is more compatible with Bennett's understanding of "popular heroes," Brooker's understanding of "cultural icons," and even Wilde's understanding of "meta-narrative characters." No less importantly, though, I prefer Denson and Mayer's terminology because it highlights the conceptual difference between characters as represented entities with an intentional inner life that are located in storyworlds, on the one hand, and the complex cultural constructs that may eventually arise from our contemporary media culture's tendency to adapt, expand, and modify previously represented characters across the borders of both individual media texts and conventionally distinct media. While I readily acknowledge the cultural saliency of what could then, perhaps, be called transmedia figures (or "popular heroes," or "cultural icons"), the following primarily aims at unpacking some of the complexities surrounding transmedia characters in a rather more narrow sense. 7 me that the complex cultural constructs in question should not be conceptualized as merely additive, as "the sum" of all locally represented characters sharing the same name. 7 To me, the main reason for such a terminological distinction (i. e., between transmedia characters and transmedia figures) would be to reduce the conceptual confusion around the term "character," but Denson and Mayer also convincingly argue that their "terminological distinction of characters versus figures is [...] not arbitrary. A character connotes depth and complexity, while the figure in its flatness must be considered in close relation to a background, i. e., its narrative or medial horizon" (2018: 70). Evidently, this also once more connects to Wilde's observations about the "pre-narra-

Constructing characters
Let me start, however, by offering a slightly more detailed account of how what we could call local work-specific characters are constructed based on individual media texts. 8 While there certainly are different ways to describe these construction processes and the kinds of constructs that result from them, it would seem to me that Eder's conceptualization of characters as intersubjective communicative constructs "based on normative abstractions about ideal character-imaginations" (2008a: 78; "beruhen auf normativen Abstraktionen über ideale Figurenmodelle") allows for a sophisticated description of characters as opposed to their medial representation by (usually) narrative media 9 and the mental representations that the recipients of these media construct of the characters that are thus represented. Indeed, Eder repeatedly emphasizes the importance not to conflate the medial representations of characters or the individual imaginations that recipients form based on these medial representations with the characters themselves. 10 Further following this line of reasoning, any discussion of how characters are represented or, rather, intersubjectively constructed across conventionally distinct media such as literary texts, comics, films, television series, or video games needs to take into account recipients' collective mental dispositions, medium-as well as genre-specific communicative rules or representational conventions, and hypothetical authorial intentions. 11 tive" qualities of the kyara, whichin contradistinction to the kyarakutāmight then be more appropriately understand as a "figure" instead of a "character." It would go beyond the scope of this article to discuss the resulting processes of meaning making (or, indeed, their similarities and differences with regard to conventionally distinct media forms) in too much detail, but I would at least like to mention two complementary principles that have been primarily developed with represented worlds in mind, but turn out to be no less relevant with regard to the intersubjective construction of characters as entities located in these (story) worlds. According to Marie-Laure Ryan, there is a principle of minimal departure at work during narrative meaning making that has as its object various kinds of represented worlds or storyworlds. This principle of minimal departure allows the recipients to "project upon these worlds everything [they] know about reality, [making] only the adjustments dictated by the text" (Ryan 1991: 51). Similar observations can be found not only in various other theories of representation but also in theories of character, with Fotis Jannidis, for example, distinguishing between three forms of knowledge that recipients draw on when imagining characters across media "(a) the basic type, which provides a very fundamental structure for those entities which are seen as sentient beings; (b) character models or types such as the femme fatale or the hard-boiled detective; (c) encyclopedic knowledge of human beings underlying inferences which contribute to the process of characterization" (2014: 30, original emphases).
It is worth stressing, though, that recipients do not complete the gaps or indeterminacies in storyworlds from the actual world itself but rather from their actual world knowledge, and that, moreover, "the frame of reference invoked by the principle of minimal departure is not the sole product of unmediated personal experience," but may include various forms of medial and generic knowledge, or even a specific "textual universe as frame of reference" (Ryan 1991: 54). The relevance of the principle of minimal departure for the issue at hand seems obvious, then, as recipients will draw not only on what Jannidis calls their "encyclopedic knowledge of human beings" but also, and sometimes quite saliently, on their knowledge about previously represented characters as well as transmedia character templates or transmedia character types. What I mean here is that the aforementioned cultural saliency of transmedia figures leads to recipients commonly having previous knowledge and/or expectations about some of the characteristics of a character that is called Sherlock Holmes, Batman, or Lara Crofteven if the fact that two characters may share the same name is arguably not sufficient for assuming that they are the same character.

Transmedia characters: Theory and analysis
A helpful way of thinking about the function that knowledge about transmedia figures plays in the intersubjective construction of work-specific characters may be offered by conceptualizing such knowledge in terms of what Roberta Pearson describes as an "established character template" (2018: 150) in the context of her analysis of current Sherlock Holmes characters. Such a transmedia character template would include physical, mental, and social characteristics of an established transmedia figure that any work-specific character sharing the same name may or may not exhibit, but would initially be expected to exhibit via a character-specific version of the principle of minimal departure. 12 No less importantly, recipients may also recognize that a work-specific character belongs to a certain transmedia character type, such as "being a space marine" (in the transmedia universe of Warhammer 40,000, or some other, similar transmedia universe; see, e. g., Baumgartner 2015); or "being an elf" (in the transmedia universe of The lord of the rings or some other, similar transmedia universe; see, e. g., Klastrup and Tosca 2011); or "being a/the great detective" (as is the case for both Sherlock Holmes and Batman characters; see, e. g., Pearson 2017a); or "being an adventurous archeologist" (as it the case for both Indiana Jones and Lara Croft characters; see, e. g., Hernández-Pérez and Ferreras Rodríguez 2014; Kennedy 2002).
I will come back to the sometimes rather complex ways in which these different kinds of knowledge may or may not inform recipients' construction of workspecific as well as transmedia characters, but first, I would like to stress again that the principle of minimal departure is not all there is to the construction of characters across media. 13 No less importantly, recipients routinely "ignore" some as-12 Transmedia character templates can thus be understood to be in some ways similar to what Lisbeth Klastrup and Susana Tosca describe as the "'worldness' (a number of distinguishing features of its universe)" of "transmedial worlds" (2004: 409, original emphases). In light of the synchronic complexity and diachronic variability of transmedia franchises, however, it is distinctly possible for more than one transmedia character template to be derived from a transmedia figure. See also Pearson 2017aPearson , 2017bPearson , 2018 for further discussions of salient characteristics and competing transmedia character templates in the context of Batman and Sherlock Holmes. 13 It may also be worth stressing that, despite the importance of the principle of minimal departure some of the gaps in medial representations of characters cannot be completed in an intersubjectively plausible manner. The main reason for this is that recipients knowledge about characters, character templates, and character types will always leave some questions unanswered and, indeed, unanswerable. While recipients may pretend that characters are complete and "human-like" beings, then, characters as represented entities will remain actually incomplete. This still holds with regard to characters that are represented beyond the borders of a single work or individual media text, though the transmedial representation of characters certainly makes it more challenging to determine whether or not an answer to any particular specific question is availableand if it seems to be available for previously represented characters with the same name, whether or not it is pects of medial representations in order to intersubjectively construct the characters that literary texts, comics, films, television series, video games, or other media forms may represent. Put simply, the intersubjective construction of characters is often based on an acute awareness of the intricacies of what Gregory Currie calls representational correspondence, a term designed to capture the general observation that, "for a given representational work, only certain features of the representation serve to represent features of the things represented" (2010: 59). Particularly in cases where the assumption of representational correspondence becomes problematic, recipients will look for alternative external explanations related to hypothetical authorial intentions or established representational conventions rather than rigidly insisting on internal explanations.
Kendall Walton pointedly describes the limits of representational correspondence in terms of a principle of charity, noting that [t]he generation of fictional truths is sometimes blocked (if not merely deemphasized) just, or primarily, because they make troublebecause they would render the fictional world uncomfortably paradoxical. If there is another ready explanation for the artist's inclusion of a feature that appears to generate a given fictional truth, it may not seem that he [or she; JNT] meant especially to have it generated. And this may argue against recognizing that it is generated. (Walton 1990: 183, original emphasis) Put in a nutshell, recipients will generally try to exhaust every possible alternative explanation before attempting to imagine "uncomfortably paradoxical" or even just comparatively "inaccessible" characters. Most saliently, these explanations will refer to medium-specific representational conventions and constraints on the production of medial representations (see, again, Thon 2015Thon , 2016Thon , 2017, and the resulting awareness of the limits of representational correspondence also applies to the transtextual or transmedial representation of characters. Most readers of Neil Gaiman's The Sandman series , for example, will understand that it would amount to what Currie and Walton would call a "silly question" (Currie 2010: 59;Walton 1990: 176) to ask why the outer appearance of the main character Dream has "changed so much," when what actually changes are the drawing styles of the various artists with whom Neil Gaiman has collaborated over the course of the comics series. Likewise, it would be "silly" to ask for an internal explanation of the differences between the audiovisual representation of Albus Dumbledore in the first two feature films of the Harry Potter series (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011) and the corresponding audiovisual representation in its subsequent installplausible to assume the answer remains valid for different work-specific characters that are called, for example, Sherlock Holmes, Batman, or Lara Croft.
Transmedia characters: Theory and analysis ments, as the untimely death of actor Richard Harris, who was then replaced by Michael Gambon, provides a ready external explanation of these differences. Finally, at least some of the rather noticeable differences in the ways that the various entries in the Tomb raider video game series (1996-present) have represented an adventurous archaeologist called Lara Croft will be charitably ignored based on widely shared knowledge about the technological limitations of graphic engines and the resulting constraints of in-game medial representations.
In the absence of an internal explanation for particularly noticeable change on the level of the storyworld, 14 some degree of apparently contradictory difference in the medial representation of "series characters" tends to be charitably ignoredand this would also seem to apply to transmedial representations of characters such as Albus Dumbledore in Joan K. Rowling's seven Harry Potter novels (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007), their eight feature film adaptations, and the various Harry Potter video games that, in turn, tend to base their audiovisual representations of Dumbledore on the way he is represented in the films; or to those of adventurous archeologist Lara Croft in the various entries of the Tomb raider video game series, the two Tomb raider feature films starring Angelina Jolie (2001,2003) and the current production starring Alicia Vikander (2018), or the various Tomb raider comics published by Top Cow (1999)(2000)(2001)(2002)(2003)(2004)(2005) and Dark Horse (2014-present). Just like there are limits to the principle of minimal departure, 15 however, there are limits to the principle of charityand while the exact threshold at which differences in the medial representation of characters cannot be charitably ignored anymore may vary from case to case (and, indeed, from recipient to recipient), we would best speak of two different work-specific characters that do not coalesce into a single transmedia character in those cases, even if these characters share 14 One of the more salient cases would be the BBC's long-running television series Doctor who, which explains the regularly occurring changes of the actor used to represent its protagonist as the result of all time lords' capability for "regeneration," a process that transforms both their physical form and some aspects of their personality and has recently led to a sex change of the eponymous character, who is currently played by Jodie Whittaker (see, e. g., Hills 2010Hills , 2015. Perhaps unsurprisingly, this most recent change proved somewhat controversial, but it seems likely that the hesitation by some "fans" of the show to accept the well-established internal explanation ("regeneration") for the change in the Doctor's physical form was perhaps less motivated by an awareness of the intricacies of narrative consistency in long-running television series than by plain old sexism (see also, again, Hills 2018; as well as Harris and Ridley 2019). 15 Walton, for example, distinguishes between the "reality principle," which would make "fictional worlds as much like the real one as the core of primary fictional truths permits" (1990: 144-145), and the "mutual belief principle," which "directs us to extrapolate so as to maximize similarities between fictional worlds and the real world not as it actually is but as it is or was mutually believed to be in the artist's society" (1990: 152). the same name, are written by the same author, drawn by the same artist, played by the same actor, etc. 16

Correlating characters
As will have become sufficiently clear by now, this article aims at developing a "bottom-up" as opposed to a "top-down" approach to the analysis of transmedia characters that initially focuses on specific instantiations of characters as they are represented in individual media texts before asking how these local work-specific characters relate to other local work-specific characters within a relevant transmedia context. Arguably, this allows us to evade any undue emphasis on the "model of the single character" that tends to dominate the current research landscape of transmedia studies, allowing us to instead acknowledge that there is no reason to assume that all represented entities sharing, for example, a common name, are or should be treated as the same character. Again, I do not at all mean to deny that what I have called transmedia figures are important and often quite visible elements of contemporary media culture or, indeed, that they make salient contributions to shaping the transmedia character templates and transmedia character types that at least partially orient how characters across media are intersubjectively constructed.
Still, I would maintain that these transmedia figures should not be conflated with transmedia characters, not least because any approach to the analysis of the latter should be able to acknowledge what Henry Jenkins has described as "a moment of transition from continuity to multiplicity" (2009: 22) in many transmedia franchises. On the one hand, then, our terminology should be able to analyze the consequences of this "shift away from focusing primarily on building up continuity within the fictional universe and towards the development of multiple and contradictory versions of the same characters functioning as it were in parallel universes" (Jenkins 2007: n.pag.). On the other hand, we should also be able to analyze those cases where "continuity" is employed in order to represent a single character across the borders of individual media texts and conventionally distinct media. While the proposed conceptualization of characters as represented entities 16 Vice versa, being written by the same individual author, being drawn by the same artist, or being played by the same actor is of course not a strict requirement for two work-specific characters to coalesce into a single transmedia character. While this may still be comparatively rare within a single work or series of works (see Thon 2017 for discussion of examples), insisting that medial representations of a character have to be produced by a single author or author collective (even in a very broad sense) would underestimate the power of the principle of charity.
with an intentional inner life that are located in storyworlds certainly allows for such glocal transmedia characters, it ultimately questions the existence of, say, a single Sherlock Holmes character, Batman character, or Lara Croft character. Rather, in all of these cases, it seems more appropriate to emphasize these characters multiplicity by talking about global transmedia character networks consisting of work-specific characters, some of which may, under certain conditions, coalesce into a single transmedia characters.
What are the conditions, then, under which we can talk about a transmedia character within the framework that I have sketched thus far? More generally, how can we analyze the relations between work-specific characters that are part of a particular transmedia character network? Drawing on Marie-Laure Ryan's discussion of "expansion" and "modification" (2008: 385), Henry Jenkins's discussion of "adaptation and extension" (2011: n.pag.), and Mark J. P. Wolf's discussion of "growth and adaptation" (2012: 245), I have previously suggested that two single narrative works within a transmedia franchise can be defined, first, by a relation of redundancy, when one is aiming to represent the same elements of a storyworld that the other represents; second, by a relation of expansion, when one is aiming to represent the same storyworld that the other represents, but adds previously unrepresented elements; and, third, by a relation of modification, when one is aiming to represent elements of the storyworld represented by the other, but adds previously unrepresented elements that make it impossible to comprehend the elements represented by the two narrative works as part of a single, noncontradictory storyworld (see Thon 2015). Keeping in mind that I conceptualize characters as medially represented entities with an intentional inner life that are located in storyworlds, these are also the three kinds of relations between work-specific characters in a transmedia character network that I will distinguish in the following.
Particularly in the context of transmedia franchises that are governed by "multiplicity" rather than "continuity," modification will often be the dominant relation between work-specific characters. Indeed, there can be little doubt that the Victorian master detective Sherlock Holmes represented by Sir Arthur Conan Doyle's short stories and novels , the 21st-century Sherlock Holmes represented by BBC's Sherlock (2010, and the 21st-century American immigrant Sherlock Holmes represented by CBS's Elementary (2012-present), the 21stcentury African-American Sherlock Holmes in Boller, Leonardi, and Stroman's comics series Watson and Holmes (2013-present), the canine master detective in the Italian-Japanese anime series Sherlock hound (1984)(1985), or the rodent master detective in Walt Disney's animated film The great mouse detective (1986) do notand do not seem to be intended tocoalesce into a single transmedia character. Likewise, the differences between, say, the Batman represented in the early Detective comics of the 1930s and 1940s, the Batman played by Adam West in the Batman television series (1966)(1967)(1968), the Batman represented in Batman: The animated series (1992)(1993)(1994)(1995), and the Batman played by Christian Bale in the three films of Christopher Nolan's The dark knight trilogy (2005,2008,2012) are rather noticeable and treating all of these Batmen as a single character would go beyond what even the most charitable recipient may be willing to tolerateand, indeed, would go against DC Comics' emphasis on the "mainstream continuity of the comic book" as presenting the most "canonical" version of Batman, though it is worth noting that "even the Batman [or, rather, Batmen; JNT] of contemporary comic books [are] far from a unified, coherent character" (Brooker 2012: 77), despite sharing the same name.
It seems worth noting at this point that some literary theorists have identified authorial continuity as a salient requirement for different media texts to contribute to the representation of a single transtextual or transmedia character. Ryan, for example, suggests that "all autonomous, self-sufficient works with distinct authors (or groups of authors) have different fictional worlds, although of course the converse does not hold: distinct works by the same author may or may not refer to the same world" (2008: 393; see also Doležel 1998;Saint-Gelais 2011). Brian Richardson likewise discusses the requirement of authorial continuity, but also acknowledges that, "[i]f the criterion for continuous identity across texts is authorial designation (tempered by consistency and, when appropriate, mimetic fidelity), then authors may equally appoint others to extend their created worlds" (2010: 533) and, thus, expand their previously represented characters. While it would seem misguided to discard authorial continuity as a requirement for strong forms of narrative continuity, then, it seems equally clear that the relations between work-specific characters within transmedia character networks is seldom governed by the authority of single authors, but rather by multiple authors or author collectives with varying degrees of authority as well as by the kinds of institutional authorship that often takes the form of IP ownership in current media culture (see also, e. g., Gray and Johnson 2013;Johnson 2013).
Depending on the specific authorial configuration of a franchise, some authors (of sorts) will have the authority to declare that certain work-specific characters are meant to coalesce into a single transmedia character, while others will be more limited in the claims they can make about the supposed expansion of previously represented characters. As Wolf explains, authorship can thus be conceptualized as a series of concentric circles extending out from the world's originator (or originators), with each circle of delegated authority being further removed from the world's origination and involving diminishing authorial contributions, from the originator and main author to estates, heirs, and torchbearers; employees and freelancers; the makers of approved, derivative, and ancillary products that are based on a world; and finally to the noncanonical additions of elaborationists and fan productions. (2012: 269)

Transmedia characters: Theory and analysis
Colin B. Harvey describes the representational constraints resulting from the increasingly complex authorial configurations of transmedia franchises as a form of "legally proscribed memory," stressing that "legally binding documents [...] dictate what elements of a franchise can and cannot be used and in what context" (2014: 279). While it is clear that the relations between work-specific characters are subject to powerful normative discourses that draw on the authority that comes with authorship or IP ownership, then, the degree to which transmedia character networks are policed by authors and/or IP owners varies, and different authorial agents may strive to control certain parts of the network, but not others.
In the case of the Sherlock Holmes transmedia character network, for example, Conan Doyle initially "exercised no active control over the screen adaptations produced during his lifetime" and "[h]is descendants exhibited an even greater desire to exploit the property and an even greater indifference to 'fidelity' to their father's work" (Pearson 2017b: 118). Now that much of Doyle's works have entered the public domain, 17 the work-specific characters within the Sherlock Holmes transmedia character network will likely have become only more diverse, but one can still find media texts that contribute to the representation of a single transmedia character across media boundaries. That being said, the focus of licensed tiein products that explicitly relate to some of the more visible current work-specific characters in the Sherlock Holmes transmedia character network tends to be on redundancy rather than expansionas illustrated, for example, by the Sherlock manga series (2012-present), the mobile app Sherlock. The network (2014), and various BBC-produced and authorized websites such as http://www.johnwatson blog.co.uk (accessed: 1 July 2019).
As Pearson has reconstructed in more detail, however, even in the absence of a single dominant transmedia character or an author/IP owner that would be in a position to establish "canonical hierarchies" ex cathedra, fans have discussed in painstaking detail how, for example, the Sherlock Holmes represented by the BBC's Sherlock or the Sherlock Holmes represented by CBS's Elementary relate to the previously represented work-specific characters in the Sherlock Holmes transmedia character network. As Pearson puts it, 17 As Pearson notes, "the character elements introduced in the few remaining stories still under copyright themselves remain under copyright. For example, anyone producing a text featuring Watson's second wife must seek permission from the CDE [Conan Doyle Estate] until 2023" (2017b. It is also worth stressing here that the copyright of many media texts representing work-specific characters called Sherlock Holmes is still active, and some of them (such as the BBC's Sherlock or CBS's Elementary) can be considered "mini-franchises" in their own right.
[g]enerally, critics, reviewers and bloggers focused on the character's psychological traits and habitual behaviours, rather than upon demographics, physical attributes, secondary characters and geographic settings. [...] While no consensus emerged on these and other questions, viewers' evaluations were predicated primarily upon the fit between the [Cumberbatch-Holmes]'s and [Miller-Holmes]'s psychological traits and habitual behaviours and those of [...] the sum of all previous incarnations of the Great Detective. (2018: 154) Even when medial representations of characters are clearly not meant to contribute to the representation of a single, consistent transmedia character, then, reception is evidently still informed and oriented by previously established transmedia character templates in rather consequential ways.
Yet, this kind of "multiplicity" is certainly not limited to transmedia character networks that emerge from franchises without a central IP owner. On the contrary, the synchronic as well as diachronic complexity of long-running franchises often also complicates the surrounding normative discourses. As Brooker notes, discourses around the real Batman, the original Batman, the Batman faithful to 'the source', persist: they are circulated, shared and reinforced by fans, authors, artists, journalists and editors. These conversations debateand depending on the cultural power of the person or organization involved, decidewhich Batman, or Batmen, are official, in continuity and canon. They decide how the official versions of Batman relate to each other. They decide which type of Batman is the current dominant, and which one is aberrant. The approved Batman is promoted, and the Batman that doesn't fit is pushed aside. (2012: xi-xii) Of course, the observation that these normative discourses exist and that institutional or other kinds of authors regularly use "explicit editorial statements about the canonical and the non-canonical" (Pearson and Uricchio 2015: 214) to paratextually shape the relations between work-specific characters within transmedia character networks should not be mistaken for a normative judgment on the "value" of any particular (type of) work-specific character. While I would maintain that it remains important to acknowledge that the intersubjective construction of work-specific as well as transmedia characters tends to be oriented by hypothetical authorial intentions and that expanding a previously represented character is thus usually taken to require some kind of "authorization" to do so 18 (with "unauthorized" expansions perhaps best thought of as drawing on previously represented work-specific or transmedia characters, but allowing for the transfer of 18 Indeed, the kind of "editorial statements" that may provide such "authorization" are effective mainly because recipients perceive them as "authoritative." As, for example, Brooker notes with regard to the Batman franchise, "[c]ontemporary continuity is policed in more detail by fans, who document each official event in Batman's fictional career on both Wikipedia and the more specialist DC Database, and explain it on forums like Comic Book Resources" (2012: 154). character-specific knowledge in one direction only 19 ), it should be clear by now that a transmedia character network includes all work-specific characters sharing the same name (or being otherwise identifiable as belonging to the network in question) and that, from the perspective of the framework presented here, modifying a character does not require any "authorization" whatsoever.
Accordingly, even character-oriented transmedia franchises whose IP owners keenly attempt to "police multiplicity" (Pearson 2017b: 121) are usually unable to make all of the various medial representation of their often eponymous main character coalesce into a single transmedia character. The differences between work-specific characters may be less immediately noticeable in the case of the Tomb raider transmedia character network then they are in the case of the Sherlock Holmes or the Batman transmedia character networks, for example, but some differences between the various medial representations of work-specific characters called Lara Croft are still too noticeable to be charitably ignored or reconciled. This has let the current IP owner, Crystal Dynamics, to "reboot" the franchise with the publication of the video game Tomb raider: A survivor is born (2013), establishing a new "canonical" character that has since been claimed to have been transtextually and transmedially expanded by Rhianna Pratchett's prequel comic Tomb raider: The beginning (2013), Dan Abnett's and Nik Vincent's novel Tomb raider: The ten thousand immortals (2014), and the video game sequels Rise of the tomb raider (2015), and Shadow of the tomb raider (2018). 20 On the one hand, then, it seems clear that the IP owner of a franchise can declare a "reboot" and thus start a new line of continuity that, by virtue of the authority that comes from IP ownership (rather than authorship in a more narrow sense), establishes a new cluster of work-specific characters that may be intended to coalesce into a single transmedia character. On the other hand, though, the relations between work-specific characters remain the object of complex normative discourses, and while it seems that recipients generally accept that the recent Tomb raider video games as well as Dark Horse's "rebooted" comics series  In franchises with multiple well-defined canons (such as A Song of Ice and Fire/Game of Thrones, The Vampire Diaries, or The Walking Dead), fan fiction authors often make this "one-way relation" explicit by specifying the body of "authorized" work that they draw on, while also stressing that they do not "own" the characters represented in these works. For further discussion of how fan fiction complicates traditional notions of authorship and canonicity, see also, e. g., Busse 2017; Fathallah 2017; Lindgren Leavenworth and Isakkson 2013. 20 By now, the "reboot" has become an established practice and the term is used in both production and reception discourses. As Wolf notes, "[t]he majority of the time, reboots appear in character-based franchises; they are done to update long-running franchises which have become dated over time, and they are usually done by people other than the original creators of the franchise (which naturally leads to discussions of canonicity)" (2012: 215). present) contribute to the representation of a single transmedia character, Crystal Dynamics claim that Tomb raider: The ten thousand immortals should also be considered to be part of the "rebooted" Tomb raider canon does not seem to be generally accepted by fans, who note that the novel contradicts various events that were represented in the comics (see Tomb raider wiki 2018). For similar reasons, and in the absence of "explicit editorial statements" to the contrary, the 2018 Tomb raider film starring Alicia Vikander is generally considered to have "rebooted" a separate "film canon," rather than expanding any of the previously represented work-specific characters called Lara Croft (see Wiki raider 2018).

Conclusion
In conclusion, let me stress again that much more could be said on many of the theoretical issues that the present article has tried to address as well as on the three transmedia franchises 21 that I have mainly used to illustrate my argument. While I certainly acknowledge the importance of what I have called transmedia figures, however, the proposed theoretical framework privileges the analysis of characters that are represented in single works or individual media texts, before asking how these local work-specific characters relate to other local work-specific characters or coalesce into glocal transmedia characters as part of global transmedia character networks, thus acknowledging that there really is no reason to assume that medial representations of characters sharing the same name are representations of the same character. The connections between these work-specific characters within a transmedia character network could then be described as either relations of redundancy, relations of expansion, or relations of modificationwith only redundancy and expansion allowing for medial representations of work-specific characters to contribute to the representation of a single transmedia character. 22 In intersubjectively constructing characters across media, how-21 As has already been mentioned, the Sherlock Holmes transmedia character network as well as the Batman character network and the Lara Croft character network are not completely controlled by a single IP owner, but while the latter may be transmedia franchises in a more traditional sense than the former, it still seems plausible to use the term in all three cases as "a cultural shorthand for understanding the expansion of cultural production across different media and industry sectors" (Johnson 2013: 27). See also Pearson 2015 for further discussion. 22 It may also be worth noting here that the ways in which individual recipients imagine workspecific as well as transmedia characters (or transmedia character networks and transmedia figures, for that matter) can differ substantially, depending on the media texts they have previously encountered and the order in which they have encountered them in. As, for example, Wolf notes, "[a]lthough a series of works can be experienced in any order, there are six types of orderings that ever, recipients will also draw on their previously accumulated knowledge about work-specific or transmedia characters as well as about transmedia character templates and even more general transmedia character typesand, no less importantly, the relations between work-specific characters within transmedia character networks are governed by powerful normative discourses that police (or at least attempt to police) whether or not a given media text has the "authority" to expand (rather than merely modify) a previously represented character. 23 As my brief discussion of the Sherlock Holmes, Batman, and Tomb raider franchises will have hinted at already, the ways in which the relations between work-specific characters within transmedia character networks are negotiated via such normative discourses may differ substantially from case to case, requiring in-depth engagement with an often quite overwhelming wealth of material that goes significantly beyond the literary texts, comics, films, television series, video games, or other media forms that may be used to represent the characters in question. Yet, even if some of the theoretical issues arising from the synchronic as well as diachronic complexity of transmedia character networks and their relation to the no less complex cultural constructs that I have provisionally called transmedia figures will have to remain unexplored for now, I am cautiously optimistic that the perspective on the theory and analysis of characters across media presented throughout the preceding pages will have at least some heuristic value for future studies of one of the most salient elements of contemporary media culture.
are most likely to occur, each of which changes one's experience of a world: order of public appearance, order of creation, internal chronological order, canonical order, order of media preference, and age-appropriate order" (2012: 264-265). While not the central concern of this article, it would certainly be instructive to explore how actual recipients' individual reception histories may or may not influence the ways in which they understand the relations between work-specific characters in transmedia character networks (see also Klastrup and Tosca 2014). 23 Yet, while Wolf may be right in remarking that "for a work to be canonical requires that it be declared as such by someone with the authority to do so" (2012: 271), my brief examination of the normative discourses surrounding the recent Tomb raider "reboot" will have illustrated already that the power of "editorial statements" by the authors or IP owners of a franchise is often quite limited and that the question of "canonicity" mainly becomes relevant in those cases where the medial representation of a work-specific character can be comprehended as a noncontradictory expansion rather than as a contradictory modification of a previously represented work-specific or transmedia character within a given franchise's transmedia character network.