TY - JOUR AU1 - Murel,, Jacob AB - Abstract Building upon Walsh’s Comic Book Markup Language (CMBL) used for encoding text features of comics documents, this essay explores how CBML can be modified and expanded using additional Text Encoding Initiative (TEI) features to reflect alternative theoretical and critical approaches to comics. In doing so, this essay argues that markup languages offer not only a means for analyzing encoded documents but also a means for analyzing critical approaches to documents. Because markup language reflects the critical stance of whoever produces the encoding, any revision to the markup potentially reflects a revision to the critical theoretical framework from which the encoder operates. As such, implementation of markup language in comics studies can function not only as a metalanguage for describing comics but also as a form of meta-criticism. To this end, this essay explores methods for incorporating CBML and TEI to reflect commonly opposed approaches to analyzing comics documents. 1 Introduction In a 2012 issue of Digital Humanities Quarterly, John Walsh introduces a Text Encoding Initiative (TEI)-based Comic Book Markup Language (CBML) for digitally transcribing comics. As markup languages, both TEI and CBML offer an XML vocabulary of machine-readable tags used to identify, query, and analyze the structural and semantic features of an encoded document. In other words, as a markup language, CBML operates as a digital metalanguage used to describe the bibliographic structure of documents widely categorized as ‘comics’. One purpose behind developing a markup language specifically for comics, is, as Walsh writes, ‘to support the study and analysis of comic books in the way that digital collections of, for instance, English poetry or American fiction support the study and analysis of more traditional literary forms’ (Walsh, 2012). He continues to explain: A large corpus of digitized comic books, along with encoded transcriptions and descriptive metadata, would allow scholars to search the text of comic books, search for keywords related to topics of interest, search for the appearance of particular characters, or search for works by particular writers and artists. Additionally, when exploited to its full potential, a large CBML collection would allow searching—and other forms of computer processing and computational analysis—based on structural, aesthetic, and informational and documentary features peculiar to the genre of comic books. (Walsh, 2012) Here, consistent with TEI markup’s typical textual focus, Walsh’s CBML prioritizes comics’ bibliographical and textual features, and as such, reflects only one critical approach to the comics medium. Building upon Walsh’s initial development of CBML, this essay explores how CBML can be expanded using additional features of TEI to reflect alternative theoretical and critical approaches to comics. In doing so, this essay argues that markup languages offer not only a means for analyzing encoded documents (as argued in markup scholarship) but also a means for analyzing critical approaches to documents. Because markup language reflects the critical stance of whoever produces the encoding, any revision to the markup potentially reflects a revision to the critical theoretical framework from which the encoder operates. As such, implementation of markup language in comics studies can function as both a metalanguage for describing comics and as a form of meta-criticism, that is, a way to analyze and understand various critical and theoretical frameworks for analyzing comics. 2 An Introduction to TEI Markup Language Before moving on to comics and how markup languages may be applied to comics criticism, a brief introduction to encoding and markup languages may be warranted, so as not to alienate less familiar readers. Among the humanities and social sciences, perhaps the most widely used markup language is the TEI, a well-developed model for the digital representation of various document types—e.g. manuscripts, print, prose, verse, transcriptions, etc.—developed and maintained by an international consortium of research institutions, projects, and individuals. The TEI Guidelines make recommendations about suitable ways of representing those features of textual resources which need to be identified explicitly in order to facilitate processing by computer programs. In particular, they specify a set of markers (or tags) which may be inserted in the electronic representation of the text, in order to mark the text structure and other features of interest. (TEI, 2020; emphasis in original) These ‘features of interest’ range from names and dates, to links between text documents, to individual words or parts of speech. Due to divergent disciplinary and research interests, encoding practices and features of interests will vary; a book historian may use TEI for manuscript description and bibliographical analysis, while a linguist may use it for linguistic analysis via encoding principal parts of speech or variance between speech patterns. TEI’s range of applications for encoding even one document illustrates the interpretational and subjective nature of markup languages—as Jerome McGann writes, ‘When you mark up a text you are ipso facto reading and interpreting it’ (McGann, 2001). Encoding produces not only a description of the document(s) in question but an interpretation as well. On this front, Julia Flanders writes, ‘[W]e can understand XML [and by extension its derivative TEI] as a way of expressing perspectival understandings of the text: not as a way of capturing what is timeless and essential but as a way of inscribing our own changeable will on the text—in other words, as a form of reading’ (Flanders, 2005). Flanders here conceives of the encoded document as producing a scholarly reading of the source document, much like the traditional research essay. The difference between encoding and traditional readings of texts is that ‘the source text and the scholar’s reading and analysis are inextricably intermingled in the encoded document’ (Walsh, 2012). Johanna Drucker understands encoding’s interpretational nature, its status as a form of scholarly reading, to be one of the practice’s chief merits. She describes encoded documents as metadocuments, which describe and enhance information but also serve as performative instruments. … Metalanguages have a fascinating power, carrying a suggestion of higher order capabilities. As texts that describe a language, naming and articulating its structures, forms, and functions they seem to trump languages that are used merely for composition or expression. (Drucker, 2009) As a metalanguage designed to describe bibliographic and semantic features, markup language allows for ‘a dynamic engagement with texts and not a program aimed at discovering the objectively constitutive features of what a text “is”’(McGann, 2002). Rather than merely observing the source document, be it a medieval manuscript or contemporary comic book, markup language allows for dynamic engagement with a document as the encoder dissects its numerous structural and rhetorical properties to produce an interpretational description in the form of markup. In this way, markup language provides a digital alternative to traditional methods of criticism in scholarly discourse. 3 Markup’s Relevance to Comics Criticism Comics criticism can benefit from markup languages like CBML for many reasons, the first being markup scholarship's open embrace of localized interpretation. Digital humanists such as Flanders and McGann approach markup languages as a way to embrace the localized nature of documents rather than attempting to unveil their timeless, essential natures. Much past comics criticism has not approached the comics medium this way. Since Scott McCloud’s Understanding Comics (1993), comics scholars have sought to provide a definition of the comics medium that accounts for its widely variant cultural forms by pointing to the medium’s essential nature (Eisner, 1985; McCloud, 1993; Hayman and Pratt, 2005; Chute, 2008), although, admittedly, this trend has dissipated or met with resistance of late (Cohn, 2005; Lefevre, 2010). As illustrated by McCloud’s own evident difficulty in crafting a definition to fit every variation of comics, the greatest and most obvious challenge to crafting one definition of comics, is that comics ‘are defiantly resistant to any medium specificity’, forever ‘defying any singular identity confined to their specific history in mass print media, while simultaneously remembering that history, keeping it alive in the homonym, the soundpun of comics/co-mix’ (Mitchell, 2014). Indeed, with the advent of digital and ‘motion’ comics, comics cannot even be confined to still, print media. Neil Cohn writes, ‘Ultimately, the definition of comics includes a network of ideas composed of their subject matter, format, readership, history, industry, the specific visual languages they use, and other cultural characteristics’ (Cohn, 2013b). Given this multifaceted determination of what comes to be considered a comic, comics criticism is better served by embracing the medium’s localized nature, dynamically engaging with comics, both individually and collectively. Jerome McGann’s above quote on text encoding can be applied to comics: ‘what is needed is a dynamic engagement with [comics] and not a program aimed at discovering the objectively constitutive features of what [a comic] “is”’ (McGann, 2002). Markup language offers to be one means of providing this sort of dynamic, localized engagement. Markup language additionally provides an alternative mode of criticism in comics studies, a field previously dominated by literary analysis. Hannah Miodrag critiques the prominence of literary criticism and semiotics in comics scholarship and, as one solution, explores an art-critical method of comics criticism (Miodrag, 2013). CBML offers another critical alternative, one based in digital technology. Walsh writes, ‘Perhaps the most significant motivation behind the development of CBML is the wish to support for comic books and related documents the interpretive strategies and inscriptions of readings … that occur when a text is encoded’ (Walsh, 2012). Developing a markup language specifically tailored to comics’ unique features, CBML enables another form of reading comics outside of dominant avenues. Of course, there exist other methods of digital comics analysis; Lev Manovich’s essay ‘How to Compare One Million Images?’ demonstrates another mode of digital analysis, utilizing computer vision and content-based image retrieval (CBIR) software to compare visual features across large-scale image sets (Manovich, 2012). For those in the humanities, however, Manovich’s CBIR method can be difficult to implement as reproducing such automated mass image comparison requires an advanced understanding of imaging software and computer vision. By comparison, markup languages can be more easily taught to humanities researchers (and students) as the basic use of markup languages does not require advanced technical training. In this way, CBML offers to be one viable, digital alternative to dominant modes of comics criticism for scholars and students in the humanities. 4 John Walsh’s CBML Walsh initially frames CBML’s textual focus as a response to Scott McCloud’s apparent focus on image over text in Understanding Comics. Walsh writes, ‘A possible objection to McCloud’s definition [of comics] is its subordination of text to image. In fact, McCloud’s definition makes no mention of text’ (Walsh, 2012). CBML aims to correct this oversight: ‘The ability to identify and characterize textual components of comics art and to describe and analyze the interplay between text and image is one of the chief aims of CBML’ (Walsh, 2012). Throughout his essay, Walsh exemplifies avenues for encoding the text that appears in comics documents, whether it be in the form of word balloons, narrative captions, text within the comics narrative (e.g. the Daily Bugle in Spider-man comics), and even the advertisements that accompany nearly every comic book serial. Later in his essay, Walsh provides an additional reason for his focus on comics text: the text format of markup language. He writes, ‘Comics are a visual, graphic art form combining text and image. CBML/TEI/XML is a text format. While one could certainly use CBML to describe details about any or all of the pictures in a comic book publication, such an effort would undermine the hybrid form of the comic book’ (Walsh, 2012). Here, Walsh aptly notes that images and text are two different communicative entities that operate according to diverging functional rules. As such, no matter how hard one may try, there will always remain aspects of an image that cannot be reflected in text and vice-versa. In response to this reality, Walsh focuses on how markup language (a text format) can represent text in comics. Yet there remain non-textual features of comics that may be represented in markup language, some features that even Walsh attempts to represent. One such feature is the graphic syntax of comics, that is, the lexical sign system widely utilized throughout comics texts. Scott McCloud has termed this comics’ ‘visual vocabulary’ (McCloud, 1993), while Neil Cohn more technical classification ‘closed-class items of the visual language lexicon’ (Cohn, 2013b). Word balloons are one example of this lexicon, others being the hearts that replace eyes to signify that a character is in love or the stars that float above a character’s head to signify dizziness. Walsh touches upon these iconical aspects of comics when he proposes a method for representing speech balloons via TEI and CBML markup. He creates an original TEI element that is meant to contain any text that appears within a given balloon. Not only does he assign the balloon element an @who attribute to represent to whom the balloon is connected in the comics document, but he also assigns the balloon @rendition and @type attributes to indicate the visual form and type of balloon, respectively, these two features often connected in comics, e.g. a balloon containing dialog that is shouted will often possess pointy edges as opposed to the customary smooth, circular shape of speech balloons. In this way, although Walsh focuses primarily on textual feature of comics, he takes part in a mildly semiotic approach to comics that could potentially provide a means for representing comics images in markup through the notion of a visual language. Moreover, Walsh attempts to represent in TEI and CBML markup, not only the appearance but the function and structure of comics panels, the fundamental unit of the comics medium, via his creation of an original element. Despite Walsh’s critical position contra McCloud, he does illustrate how his original can be employed to represent the function and structure of comics panels according to McCloud’s own comics theory. As Walsh writes, ‘CBML provides the element to represent the basic structural unit of comics. is a modification of TEI’s
element, which represents a generic subdivision of text in the TEI model’ (Walsh, 2012).1 Walsh provides the following example encoding of a panel from Marvel’s Captain America #193 (1976): … For clarity’s sake, I have removed from this sample Walsh’s encoding of panel contents that he includes between the element’s start and end tags. Additionally, I will not go into detail on all of the ‘attributes’ (as they are called in markup discourse) that Walsh attaches to the element, for only one is of importance to the present discussion, the @ana attribute and its accompanying attribute value ‘#action-to-action’. In describing this attribute, Walsh echoes the TEI Guidelines: ‘The attribute @ana, short for analysis, is another global TEI attribute … that points to one or more elements containing interpretations of the element on which the @ana attribute appears. Typically, @ana points to one or more , short for interpretation, elements’ (Walsh, 2012; emphasis in original). CBML’s @ana attribute points to the following element (note: again, I have removed some of the XML coding from Walsh’s element for clarity): The interp elements below include vocabulary that may be used to characterize the transition from the previous panel to the current panel. The transition types listed below are defined by Scott McCloud in his Understanding Comics. As the element’s content and subsequent element’s @xml:id values reveal, this specific encoding for describing panels and their transitions overtly draws from Scott McCloud’s own critical framework and theory of the comics panel, specifically panel transitions, which emphasizes linear reading order and sequentiality in comics. For Walsh, however, this is but one means of representing comics panels and their relationships in markup. In the course of his introduction to CBML, Walsh writes, ‘Of course, the encoder/editor may choose to augment or replace McCloud’s vocabulary with any suitable typology’ (Walsh, 2012). The present article pursues this path of inquiry in addressing the question: what if a CBML encoder were to approach a comics document from an alternative theoretical framework, say, one that emphasizes back-and-forth reading or the two-dimensional spatiality of comics rather than linear sequentiality? How might this change in a critical position produce a change in the markup? To this end, I suggest below a means of implementing CBML—and with it, TEI—to encode comics in ways representing Thierry Groensteen’s comics theory, known as arthrology, and Neil Cohn’s theory of comics’ visual language. It may go without saying that there are other critical approaches capable of serving as exciting models for encoding comics documents. Kai Mikkonen’s comics narratology is perhaps one such systematic theory capable of serving as a critical framework for comics encoding (Mikkonen, 2017). In a previous article published in Digital Scholarship in the Humanities, Bateman et al. aim to produce a detailed ‘classification scheme that allows the particularly visual organization of the pages of comics and graphic novels to be captured and empirically investigated in its own right’ (Bateman et al., 2017). Readers familiar with TEI markup can perhaps imagine specific means of implementing this classification system into Walsh’s CBML, which Bateman et al. review in the first half of their article, to produce a cohesive and exhaustive approach for describing panel layouts via markup. Paul Fisher Davies’ attempt to develop a ‘framework and set of vocabulary for describing the work of comics communication’, and more specifically, ‘the points of ligature and interaction between elements of the graphic narrative text’ may similarly provide a useful classification system readily integrated with CBML for describing the linguistic functions of comics features, whether pictorial or textual (Davies, 2019). There are certainly other approaches to classifying, describing, and theorizing the functions and features of comics documents that have gone unmentioned here. There are several reasons for selecting Cohn and Groensteen as examples of alternative comics theories fruitful for demonstrating potential modifications to CBML. First, much like Scott McCloud, whose comics theory Walsh chooses as a demonstrative example, both Groensteen and Cohn serve as foundational figures for a distinct critical position—McCloud in Anglophone comics criticism; Groensteen for French criticism rooted in continental philosophy; Cohn for a linguistic, as opposed to literary or art-critical, approach. Second, both Groensteen and Cohn’s theories have been positioned against McCloud’s. This is due to Cohn’s framing his own work against principally McCloud but also Groensteen (Cohn 2013b, 2010), as well as later critics like Hannah Miodrag positioning these three theorists in opposition (Miodrag, 2013). Such critical (self-)positioning is not unwarranted. While McCloud’s approach to comics emphasizes the linear, sequential reading of panels, both Groensteen and Cohn emphasize, albeit in different ways, the importance of nonlinear relationships to comics panels. It is this shared divergence from McCloud that allows both Groensteen’s arthrology of comics and Cohn’s hierarchical structure of comics to serve as apt demonstrations for potential advancements to Walsh’s CBML and the application of markup language to comics overall. Additionally, these differences provide occasion for showcasing how markup language can represent alternative critical frameworks and different approaches to those frameworks. 5 Thierry Groensteen’s Arthrology Groensteen’s comics theory, arthrology as he calls it, emphasizes the copresence of panels on the page, holding that every panel across a comics document bears some potential relationship to each of the others. Groensteen writes, ‘Breakdown and page layout are the two fundamental operations of arthrology, on which the [concept of] braiding eventually puts the finishing touches’ (Groensteen, 2007). For Groensteen, page layout refers to the parceling of panels across the page, while breakdown refers to the division of those panels into units of narrative time. Despite Groensteen’s mention of temporal breakdown as fundamental to his theory, he later writes that there are ‘two fundamental intuitions that guide [him]: that comics are composed of interdependent images; and that these images, before knowing any other kind of relation, have the sharing of a space as their first characteristic. And, remarkably, they do not say anything other than that’ (Groensteen, 2007; emphasis in original). Here, Groensteen reveals the two most important principles of his comics theory, the spatial and inter-relational nature of panels across a comics document. His notion of braiding exemplifies this two-fold nature of the panel. Groensteen writes that ‘the ultimate signification of a comics panel does not reside in itself but in the totality of relations in the network that it maintains with interdependent panels’ (Groensteen, 2007). Groensteen’s braiding is the operation that, ‘taking account of the breakdown and the page layout, defines a series within a sequential framework’ (Groensteen, 2007; emphasis in original). It recognizes that no panel exists in isolation from the others, and that panels across the sequence, page, and potentially entire book exist in a nonlinear network of associations (iconic or otherwise) outside of linear narrative progression. In Groensteen’s own words, Braiding thus manifests into consciousness the notion that the panels of a comic constitute a network, and even a system. To the syntagmatic logic of the sequence [typified by McCloud above], it imposes another logic, the associate. Through the bias of a tele-arthrology, images that the breakdown holds at a distance, physically and contextually independent, are suddenly revealed as communicating closely, in debt to one another. (Groensteen, 2007; emphasis in original) Put more simply, and in terms more natural to markup language, braiding is the linking together of panels as individuated and fragmentary elements across a document according to nonnarrative criteria. Although Walsh never explicitly details how this sort of linking in comics might be carried out with CBML, TEI (CBML’s parent markup language) provides for such nonlinear linking across a document in ways easily applied to comics’ spatial structure as theorized by Groensteen. Groensteen’s notion of braiding derives from a comic’s page layout, the latter which Groensteen conceives as a physical manifestation on the page, or ‘an improved and corrected version of’, the mental process he terms gridding (Groensteen, 2007).Groensteen writes, ‘Gridding consists of dividing the available space into a number of units or compartments’, beginning with the writer’s dividing of the work into chapters, and continuing with the illustrator’s (sometimes these are the same person) thumbnail sketches for panel layouts on each page (Groensteen, 2007). Though the process of gridding does not necessarily produce an evenly divided and literal grid pattern of panels on the comics page—as utilized throughout Alan Moore’s Watchmen, for example—Groensteen’s gridding speaks to a broadly defined tabularization of the comics page—indeed, the whole comics work. When considering Groensteen’s arthrology, Walsh’s suggestion of the TEI element
with an attribute @type of ‘panelGrp’ for representing page layouts or series2 would seem too general (Walsh, 2012). TEI provides the element
to describe any general division and subdivision of the document no matter its size or genre and so does not signify what defines the panels as a group—Do these panels appear on the same page? Do they behave as one narrative chunk? Do they share a visual or stylistic element that suggests they be collectively identified as one grouped unit? But Groensteen’s term gridding, not to mention the widely used term panel, suggests a more specific form of division in the comics document, a tabularization of the page into panels and panel units. This inferred tabularization in Groensteen’s terminology may suggest the using of the TEI element for representing his arthrology. The
element is inadequate for several reasons, but for the sake of brevity, this essay will limit itself to addressing two. First, the TEI Guidelines define
as representing ‘text displayed in tabular form, in rows and columns’ (TEI, 2020; emphasis added). The
element is designed for representing linguistic characters, be they letters or numbers, displayed in rows and columns. While such tables bear a superficial aesthetic similarity to conventional comics pages that are often divided into tabularized blocks (Fig. 1), data tables bear a significant difference in function to that of the comics page. The former organizes numerical and textual data into discrete units so as to more readily communicate what would otherwise be abstract information. In contrast, the comics page is a visual art object, the form of which is essential to how it conveys meaning, at least in Groensteen’s approach to comics. When considering Groensteen’s arthrology, the information conveyed in a comics page cannot be as easily divided from its visual formatting as perhaps can be done with a table. Groensteen’s emphasis on page layout and the networked nature of comics panels means the panels cannot be considered apart from their appearance on the surface of the comics page. It is not enough to note that the page is ‘tabularized’ by the division of panels, one must further record where and how those panels appear on the surface of the page and how the panels relate to one another. The second reason for the inadequacy of
is that not all page layouts in comics possess neat, gridded design depicted in Fig. 1. Consider Fig. 2, for instance, whose layout largely follows one of the pages drawn by Karl Kerschl for an issue of DC Comics’ Wednesday Comics series. Using TEI’s
element to encode this layout would not only involve immense difficulty in deciding how to partition each panel into discrete rows and columns but, and perhaps more importantly, would severely misrepresent the page layout. To produce an encoding consistent with Groensteen’s emphasis on the surface appearance of the comics page, the TEI and elements can be integrated with Walsh’s CBML. Fig. 1 Open in new tabDownload slide A standard grid-pattern page layout Fig. 1 Open in new tabDownload slide A standard grid-pattern page layout Fig. 2 Open in new tabDownload slide Non-grid page layout Fig. 2 Open in new tabDownload slide Non-grid page layout As described in the TEI Guidelines, the TEI element defines a ‘two-dimensional coordinate space, optionally grouping one or more graphic representations of that space, zones of interest within that space, and transcriptions of the writing within them’, while the element defines any two-dimensional area within that surface (TEI, 2020). Both elements receive the attributes @ulx, @uly, @lrx, and @lry. The numerical values attached to these attributes identify the space over which the and/or element(s) span across a coordinate plane defined by points along the plane’s x (horizontal) and y (vertical) axes. The attributes @ulx and @uly together provide the x- and y-coordinates a surface’s upper-left corner, while the @lrx and @lry attributes provide the x- and y-coordinates for that same surface’s lower right corner. In defining the surface of Fig. 2, one may encode the whole page surface as to represent the surface area of the entire page. By locating the upper left corner at (0, 0) and the lower right corner at (736, 1142) on the coordinate plane, the element indicates that the given page’s coordinate scale runs from 0 to 736 along the x-axis (horizontally) and 0 to 1142 along the y-axis (vertically).3 Using these attributes with the and elements highlights Groensteen’s emphasis on the page layout as a two-dimensional surface and can be used in conjunction with Walsh’s own element. For examining how to represent panels using this coordinate system, we can start with the upper-left-most panel of Fig. 2, which could be encoded inside the element as . This element would then be linked to a corresponding element via the @xml:id and @facs attributes, respectively. This encoding for the first panel of Fig. 2 would be as follows:           [panel content to be encoded here]    TEI provides the @facs attribute as a means to link elements to corresponding zones or regions of an image as delineated by coordinates within the encoded document. In other words, the @facs attribute is provided for the purpose of linking or elements with other non-coordinate-defined elements. The ability to link and is important because the and elements, with their incumbent coordinate points, only represent one aspect of Groensteen’s arthrology, more specifically, the synchronism of braiding. For Groensteen, the concept of braiding operates in ‘two dimensions’, the synchronic (the presence of panels on the page’s surface) and the diachronic (the reading of panels across the page) (Groensteen, 2007). By incorporating Walsh’s element with its attached narratological markers—e.g. the @n attribute for representing reading order or the @characters attribute to indicate which characters appear inside a given panel—both of braiding’s dimensions can be represented simultaneously within the same encoded document. The element and coordinate attributes represent a given panel’s spatial position and dimensions across the document’s surface, while Walsh’s with his initially proposed accompanying attributes can represent the panel’s narratological features such as reading order. The @facs attribute is then used as a pointer between these two elements in order to connect the spatial and narratological aspects of one panel and indicate that both represent the same entity in the source comics document. But while the above encoding may adequately represent rectangular spaces such as a single comics page represented by Fig. 1, the same encoding is grossly inadequate for representing the more irregular coordinate spaces formed by any one of the panels in Fig. 2. Fortunately, TEI provides a means of representing such non-rectangular zones within a coordinate space, one that requires a more detailed description of the zone (i.e. panel) in question. This is the @points attribute. Unlike the @ulx and its counterpart attributes above, @points lists a series of x, y coordinates (separated by one space) that together identify points on the line enclosing a panel’s given area. So in representing the top-left-most panel of Fig. 2, we may encode , thereby identifying the panel’s four corners within the page’s two-dimensional surface.4 Yet this encoding fails to acknowledge how the panel’s bottom edge is curved, in contrast to all of its other straight-line edges. In order to account for this, x, y coordinates along this edge would need to be encoded within the panel’s @points attribute, such as in this example: . Of course, the precise number of points encoded along this edge is left to the encoder’s discretion and largely dependent on the level of exactness the encoder aims for in representing the panel’s space. The encoding proposed so far has aimed at representing the two-part emphasis of Groensteen’s arthrology, page layout, and breakdown. By incorporating TEI’s element and coordinate-class attributes into Walsh’s CBML, the proposed encoding addresses Groensteen’s notion of the page as a two-dimensional surface divided into units called panels. From there, the encoding addresses Groensteen’s concept of breakdown—the division of the page-surface into narrative units—by assigning each an @n attribute that indicates its narratological position, or reading order, on the comics surface and/or in the comics sequence. More than TEI’s general division-type elements, the above encoding reflects Groensteen’s idea of the comics page as fragmented yet cohesive unit composed of multiple groupings of individuated yet interconnected panels across a two-dimensional surface area. But for Groensteen, ‘comics is not only an art of fragments, of scattering, of distribution; it is also an art of conjunction, of repetition, of linking together’ (Groensteen, 2007). In focusing on page layout and breakdown, the above encoding has only addressed Groensteen’s spatial emphasis, and so has risked treating each comics page in isolation from the rest of the document. But Groensteen understands comics as a fragmented network, whose surfaces and divisions are linked via myriad narrative, visual, and hypertextual resonances. This theory of comics as a network must be addressed by any encoding purporting to represent Groensteen’s arthrology. As Hannah Miodrag remarks, in conceiving of the comics document as a network of relations between panels and panel series, Groensteen’s arthrology emphasizes the ‘non-sequential iconic resonances’ and ‘myriad subtle echoes and repetitions that reverberate through a complex artwork’ (Miodrag, 2013). Such resonances and repetitions may have nothing at all to do with a comic’s narrative. This is what Groensteen speaks of when he writes, ‘The network that [panels] form is certainly an oriented network, since it is crossed by the instance of the story, but it also exists in a dechronologized mode, that of the collection, of the panoptical spread and of coexistence, considering the possibility of translinear relations and plurivectoral courses’ (Groensteen, 2007). In other words, panels are connected not only through literary or narrative motifs but through structural and visual echoes reverberating throughout the whole comic, and even potentially, across comics documents, e.g. one comic may purposefully recall the structural divisions or visual motifs of another, potentially more well-known, comic. While the TEI element provides a means of representing Groensteen’s emphasis on two-dimensional spatiality, it does not in itself provide a means for representing the reverberations and resonances that link panels or panel series and constitute an important component in understanding comics as a network. Fortunately, TEI provides other means for representing such links between structural, visual, and narrative units. One potential feature offered in TEI is the @corresp attribute. When assigned to one element, this attribute takes the value of a URI that points to another element in the encoded document so as to indicate that the former corresponds to the latter in a way defined by the encoder or host organization’s encoding documentation. Within the context of using CBML and TEI to represent Groensteen’s arthrology, the @corresp attribute can be assigned to panels to indicate visual or narrative, or even to whole elements to indicate correspondence regarding the spatial layout of select panel series or page layouts, such as if the layout of one recto page mirrors the layout of the adjacent verso page. But while valid, the use of @corresp is perhaps too simplistic given the myriad correspondences—whether narrative, visual, or bibliographic—a panel or panel series can share with others, both within and across comics documents. In other words, the @corresp attribute may indicate correspondence between panels or panel groups, but it precludes a means for categorizing the nature of this correspondence. There are several other means in TEI for representing correspondence with desired level of increased specificity. For the sake of space, however, only one possible solution will be addressed at length here, that solution being the incorporation of TEI’s and elements. The element entry in the TEI Guidelines states that this element ‘defines an association or hypertextual link among elements or passages, of some type not more precisely specifiable by other elements’ (TEI, 2020). Using this element, two or more features of the comics document are linked via their assigned @xml:id values with the nature of that connection defined by an @type and possibly @subtype attribute attached to the element. For example, imagine the two layouts displayed in Fig. 3 are the first (A) and last (B) pages of a thirty-two-page comic book. Each page’s bottom panel content is iconically similar in that two panels depict the same landscape from the same perspective, only the first page captures this landscape during the day, while the last page is at night. Between the two pages depicted in Fig. 3, then, there are two correspondences to be described via the element: (1) the shared page layout and (2) the bottom panels’ iconic resonance. Beginning with the shared page layout, each element corresponding to the two pages would receive an @xml:id value by which to identify it. Once these have been assigned, the element can be inserted into the document—per the TEI Guidelines, its location within the document ‘has no significance’ (TEI, 2020)—and assigned an @target attribute that points to both pages using their respective @xml:id values. Once the element and its pointers have been created, we can assign it @type and @subtype attributes to categorize the nature of the identified correspondence. Fig. 3 Open in new tabDownload slide Sample corresponding pages Fig. 3 Open in new tabDownload slide Sample corresponding pages Given their wide-encompassing nature, broad categories such as ‘narrative’, ‘visual’, and ‘bibliographic’ may suffice for the @type attribute. The primary issue with this sort of division between categories when considering page layout is that Groensteen among others recognizes the potentiality for page layout to simultaneously harbor narrative, visual, and bibliographical functions (Groensteen 2007, 2013; Peeters, 2007; Witek, 2009). In fact, one may argue this is true for many, if not most or all, of the correspondences readers might experience within a comics document. For instance, the bottom panels’ iconic resonance is not only a visual correspondence, but may also suggest narrative implications depending on the story these two pages bracket. Due to problems in classifying correspondences within and across comics as purely narrative, visual, or bibliographic, @type values of ‘layout’ and ‘content’ will be here used for the correspondences depicted in Fig. 3. The @subtype attribute can then be employed to further categorize the nature of these correspondences. The encoding for the two pages displayed in Fig. 3 would appear as follows (note: for clarity, I refrain from including the element and coordinate attributes in this example): … … … Here, @id="323" the first element indicates that the elements identified as ‘page01’ and ‘page32’ correspond to one another in page layout. The second element indicates that the elements identified as ‘panel07-01’ and ‘panel07-32’ correspond to one another via the iconographic resonance in their depicted landscapes. The @subtype value here aims to clarify, more specifically the element’s @type value but still in somewhat broadly applicable terms, how the two panels correspond to one another in content. In another scenario, were the bottom panel of page thirty-two a mirror image of the bottom panel on page one, with the sun and mountain range appearing on the right side of the panel rather than left, the attribute values @type="content" and @subtype="mirror" could be used. Readers should not get too caught up in potential @type and @subtype values at the moment, however, as the primary purpose is to demonstrate how the TEI element can be incorporated in Walsh’s CBML alongside the and elements to represent the spatialized network, what Groensteen calls the ‘spatio-topical apparatus’ of Groensteen’s arthrology. Those familiar with TEI will note that, given the potentially limitless correspondences in content and structure extant within any one comics document, encoders may find it better to eschew the use of @type and @subtype on the attribute and instead position all elements of the same @type attribute under one element. For instance, returning to Fig. 3, in the imaginary comic from which these two pages have been lifted, perhaps, in addition to pages one and thirty-two sharing the same layout, pages two and thirty-two share a layout, while pages three and thirty share their own layout, and so forth throughout the comic, with page layouts corresponding to one another in a loose chiastic structure. This sequence of correspondences could be represented as follows:        … Encoders could also forego the use of @type and instead use the @ana attribute and element to encode the various forms of correspondence between comics features much as described earlier in Walsh’s encoding of McCloudian panel transitions. How one elects to encode correspondences in layout or other features according to Groensteen’s arthrology depends not only upon encoding preferences and philosophies but also how the encoder interprets Groensteen. Throughout The System of Comics, Groensteen appears to value the spatial positioning of panels above their content, such as when he writes, ‘[T]hese [panels], before knowing any other kind of relation, have the sharing of space as their first characteristic. And remarkably, they do not say anything other than that’ (Groensteen, 2007; emphasis in original), or that the ‘reduction of the page to a collection of empty frames’ is ‘the most faithful representation of general theoretical model’ of comics (Groensteen, 2007). In both these excerpts, Groensteen suggests the spatial, nonlinear layout of panels is of prime importance for his theoretical approach. An encoder approaching a comics document with this interpretation of Groensteen may decide to only encode a comics’ bibliographical and formal features, such as page layout, and not encode panel content. Thus, though still employing the element, this encoder may forego the element and instead utilize the TEI @sameAs attribute. In this encoding, an @xml:id is assigned to one element and the attribute @sameAs to any other that corresponds with that ‘s layout. If an encoder is concerned only with representing page layout apart from content according to the above interpretation of Groensteen, the @sameAs attribute could be applied. But other encoders may interpret Groensteen differently as Groensteen also appears to infer or outright claim that panel arrangement (layout) and panel content (determined by breakdown) are positioned on equal footings, such as when he writes that the page layout ‘is not invented under the dictation of the breakdown, but according to a dialectic process where the two instances are mutually determined’ (Groensteen, 2007). Moreover, while Groensteen’s braiding may transpire across the spatial layout of a series (Groensteen, 2007), he also notes that it often relies upon visual echoes and iconic motifs within the content of panels (Groensteen, 2007). Encoders approaching a comic with this understanding of Groensteen will likely elect to encode both the spatialization of panels and their content. This approach invalidates the use of TEI’s @sameAs attribute, given it means the encoded element is the same as that to which it points, and so would imply the elements in question are the same in both layout and content. This is obviously untrue of the two pages in Fig. 3. The use of and to indicate layout similitude between panels or panel series interprets Groensteen as giving equal weight to both content and spatial positioning of panels.5 The use of or @sameAs, and even potentially @ana, to represent corresponding layouts demonstrates how certain approaches to encoding comics documents according to Groensteen’s comics theory reflect diverging interpretations of that theory. Whether an encoder adopts one of these encodings, or perhaps another form of markup not considered here, will be determined by how that encoder understands Groensteen’s arthrology and its emphases. In this way, the markup not only reflects a specific critical theoretical approach (here, Groensteen’s arthrology) but also a specific interpretation of that critical theoretical approach. 6 Neil Cohn’s Hierarchic Groupings Having hitherto provided general possibilities for representing Groensteen’s arthrology using TEI and CBML, I want to consider how these markup languages can be implemented for representing an alternative theory of comics and one that positions itself in opposition to both Groensteen and McCloud, the latter from whom, as mentioned earlier, Walsh draws in initially developing CBML. This alternative theory is that of comics’ visual language and grammar as theorized by cognitive scientist Neil Cohn. For Cohn, McCloud’s theory of ‘panel transitions and [Groensteen’s] arthrology are appealing as notions because the reader (or author) can directly experience them’ (Cohn, 2010). Following recent scholarship in psychology and linguistics, however, Cohn takes meaning in comics, whether narrative or nonnarrative, to be an unconscious production of the mind, arguing that the processes guiding sequential image comprehension remain inaccessible to conscious awareness. To this point, the sense in which ‘meaning’ is explored has no concern with any conscious sense of ‘artistic interpretation.’ Instead it is about the basic comprehension of sequential images—comparable to the creation of meaning by words in sentences. While we are consciously aware of (sometimes various) meanings of sentences, the unconscious processes that motivate these understandings remain inaccessible. The comprehension of sequential images is taken to work in the same way. (Cohn, 2010) Cohn’s theorization of comics’ visual language and grammar thereby turns from the conscious experience of the reader—a phenomenology of comics, so to speak, represented for Cohn by Groensteen and McCloud—and instead emphasizes ‘the mechanisms in the mind/brain that guide comprehension’ and produce meaning from a collection of images (Cohn, 2010). Much like Groensteen, Cohn seeks to move beyond a McCloudian emphasis on linearity and sequentiality in comics criticism. McCloud’s own theory of linearity and panel transitions builds from his ‘temporal mapping’ thesis, that is, the claim that panels equal narrative moments. Contra this, Cohn argues ‘Panels as units do not stand for moments or durations in fictive time, but direct attention to depictions of “event states” … from which a sense of “time” is derived’, further writing that ‘Images are just significations made meaningful through cognitively based concepts’ (Cohn, 2010; emphasis in original). As such, for Cohn, meaning in comics derives less from the moments depicted by panels than by the concepts they produce in tandem with the reader’s mind. Arguing for critical shift from panel transitions, and so the gutter space between panels, Cohn writes, ‘[T]he important focus for processing becomes the content of the panels. Indeed, individual panels must also derive their meanings from the mind of the reader, and, in some cases, that meaning relies on other panels in the sequence’ (Cohn, 2010; emphasis in original). Here, Cohn ultimately argues for the ‘relational aspects of panels to be explored without the presumption of time restrictions’, meaning that panels do not relate to one another, and so build meaning, from linear sequentiality (Cohn, 2010). Of course, Cohn and Groensteen diverge in that the latter does not totally reject McCloudian linear relationships between panels but understands them as only one type of relationship between panels, adding to it another logic of relations, that of the ‘associate’, by which he means nonlinear conceptual and spatial relationships between panels. In addition to McCloud, Cohn further critiques Groensteen’s arthrology for being a ‘broad concept’ operating on ‘loosely defined principles of connection’ and a ‘vague sense of connectedness’ that provides ‘no methods for meaningful analysis’ (Cohn, 2010). More directly, Cohn takes issue with Groensteen’s proclamation that ‘every panel exists, potentially if not actually, in relation with each of the others’ (Groensteen, 2007). The former claims that ‘such unrestrained transitions (semantic relations between individual panels) between every possible panel in a document would overload the working memory of the human mind’ (Cohn, 2010). Whatever one may think about this criqitue of Groensteen’s theory, encoding a comics document in markup language, according to Groensteen’s arthrology, and its incumbent mantra of potentially infinite relations showcases the very overload described by Cohn. To demonstrate this, let us consider how one might encode the panel series depicted in Fig. 4, whose layout and content are modeled after a segment from Veronique Tanaka’s Metronome (Tanaka, 2008), with all of its potential panel relations. First, the eight individual panels comprise one large implied panel via their shared landscape view. In the above section on Groensteen, the element corresponds to the page, so this cannot represent the implied panel. Is the encoder to wrap of the page’s elements within another to signify that they form one image? This suggests, however, that the implied panel is a literal panel on the page, rather than one implied by the unity of the smaller panels. In this case, a element connecting all of the panels on this page with, perhaps an @type attribute value of “multipanel” or “polyptych” (the latter term lifted from McCloud, 1993). Within this one series, however, are numerous potential groupings and links that can be made among the panels for which the markup would need to account. For instance, the panels in the top row form one conceptual unit via the tracking of the raven’s flight path, while the panels in the bottom row are connected by tracking the cartoon Poe’s walking path. Not only do the two rows form discrete conceptual units composed by the shared subject matter of their respective panels, but the two rows as whole units inversely relate to one another in that the raven recedes from view as Poe draws nearer the frame. The panel series can also be divided into four two-panel columns, as each column depicts a particular instant of narrative time from that landscape view. Moreover, were this panel series one segment of a larger comics document, the encoder would need to consider any number of visual echoes or motifs shared between these panels and others throughout the document, as these would establish further links between panels outside this series based on content or even form. Readers may be able to imagine other possible criteria for linking panels and panel groups in this series or another, further multiplying the potential connections and corresponding linking attributes and URIs in the encoded document. Fig. 4 Open in new tabDownload slide Panel series modeled after Tanaka, 2008 Fig. 4 Open in new tabDownload slide Panel series modeled after Tanaka, 2008 In contrast to this mass of potentially infinite relations, Cohn’s linguistic theory of comics holds that readers read panels in distinct, hierarchic groupings defined by ‘specific constraints—what in linguistics would be called a “grammar”—that differentiate the acceptable groupings from the unacceptable ones’ (Cohn, 2010). For Cohn, these groupings are defined by two principal and hierarchic factors: the panel’s depicted time and subject matter. He writes, ‘[P]anels representing the same time and character should be grouped first, followed by panels at the same time but different characters, then finally with panels in other times. The highest nodes of a tree structure should belong to different times, the lowest to the same times’ (Cohn, 2010). Here, Walsh’s TEI-derived
element can be used to encode Cohn’s hierarchic structures, so long as an added @subtype attribute specifies the panel group’s hierarchic level. For example, Cohn’s formulaic visualization of his above quote—‘Different times > Same time and different space/character (whole environment) > Same time and space/character’ (Cohn, 2010)—can be represented in CBML as such (the @subtype attribute names reflect Cohn’s short-hand terminology for panel groups):
 
  
  …   
 
This encoding represents how, in Cohn’s comics theory, panels representing the same character are joined together in the smallest group, which the reader (unconsciously) reads first, then reads in conjunction with those of different characters sharing the same environment, followed afterward by panels showing shifts in time. The containment of
s within
s represents how Cohn’s theory does not totally divide the panel groups from one another but instead dictates the order in which panel groups are read within a larger sequence. Applying this encoding to an actual panel sequence provides an opportunity to better understand Cohn’s hierarchic structure. For example, the four-panel sequence portrayed in Fig. 5, which is modelled after one of Cohn's original drawings (Cohn, 2010), contains a temporal shift between panels three and four (the first three panels depicting the same instant in time), and so all four panels would be grouped together in the highest grouping of ‘timeShift’. Within this grouping, a further subgroup is formed by all panels that depict the same time, regardless of character or physical setting. This group, ‘sharedEnvironment’, is comprised of panels one, two, and three because, although they depict multiple characters, all three panels portray the same moment in time. Panel four is excluded from this subgroup because it depicts a subsequent moment in narrative time. Beneath this second grouping, a final sub-subgroup is formed, itself comprised of all panels that depict the same character in the same moment of time. Regarding the Fig. 5 sequence, this sub-subgroup contains panels two and three as both depict the same character (i.e. Poe), and only that character, at the same moment in time. Panel one is excluded from this group because, although it depicts the same instant in narrative time as panels two and three, it depicts a different character (i.e. the raven). In accordance with Cohn’s comics theory, this hierarchic structuring of the sequence in Fig. 5 can be represented via CBML as follows:
     
     
            
 
Fig. 5 Open in new tabDownload slide Panel sequence modeled after Cohn, 2010 Fig. 5 Open in new tabDownload slide Panel sequence modeled after Cohn, 2010 That the
element as used here does not indicate a physical or visual division within the comics document is, of course, a nonissue, the TEI Guidelines stipulate that the
element ‘contains a subdivision of the front, body, or back of a text’—there is no necessary reason for this subdivision to exist on the page as a material or visual property as opposed to only within the reader’s or encoder’s mind (TEI, 2020). But while aptly representing Cohn’s theorization of how readers cognitively process a panel sequence, the above markup does not reflect the actual linear appearance or conscious reading order of panels across the page, that is, left-right. The numerical attribute values assigned to each via the @n attribute means to provide some indication of this linear appearance. Fortunately, in the above example, only panel four is encoded outside of linear reading order, although one can easily imagine a sequence in which more than one panel may reside outside of conventional linear sequence. TEI’s
element does provide an excellent avenue for understanding and representing the hierarchically structured groups characterizing Cohn’s theory of cognitive panel processing. Yet it does not provide a simultaneously apt representation for the actual appearance or conscious reading order of a given panel sequence. In contrast, the TEI element can be applied to Walsh’s CBML so as to retain the source comic’s linear reading order while still enabling panels to be grouped as Cohn describes. The TEI Guidelines indicate that the element specifies a cohesive, though fragmented, text segment by pointing to the discontiguous elements comprising it, e.g. when two or more characters each speak part of a single metrical line, specifies that the characters’ dialog, though divided by quotations and spatially on the page, comprise one metrical line. Although this fragmentation diverges from the present issue of reflecting Cohn’s cognitive groupings in markup, the element nevertheless can serve the present issue in its ability to create virtual element groups by selecting specific constituent elements from another sequence. In using , each element receives an @xml:id attribute, by which the element will target each panel and specify which panels reside in which panel group. Via its own @result attribute, then specifies what virtual (i.e. conceptual) element (here, a
element representing a panel group) is formed in grouping these together. Using , the encoding for Fig. 5 might be:
           
This encoding using the TEI element is favorable for two reasons: (1) it represents the panels as they appear in linear sequence on the page and (2) the virtuality of the panel groups theorized by Cohn—that is, that they do not exist materially on the page, nor even consciously within the reader’s mind—is represented through the element’s creation of these panel groups as virtual
elements. Yet this alternative encoding is less preferable than the previous due to the element not being able to identify the nature of each virtual
formed by the panels. In other words, by using , readers of this encoding would not know that a ‘panelGrp’ is formed by each ’s identified panels. Furthermore, the markup along does not make clear to which of Cohn’s hierarchic groupings each or virtual
corresponds. A project-specific schema customization to standard TEI, and even Walsh’s publicly available CBML schema,6 can provide a means for indicating which of Cohn’s hierarchic groups is formed by each . In the same way that Walsh creates TEI elements specifically for CBML, so technically proficient users of TEI can establish new elements as well as rename present TEI elements. Rather than a generic
element, we can create new hierarchic
-like elements akin to Cohn’s panel groups that are defined by the encoding schema so that a higher level grouping cannot appear within a lower level one. By way of a schema customization building upon Walsh’s present CBML schema, we can create elements corresponding to Cohn’s panel group names: , , and . The hierarchical nature of these attributes can be further defined in the schema so as to remain consistent with Cohn’s theoretical framework. That is, each of these elements can be defined in the encoding schema so that cannot be contained by or and cannot be contained by . Note, however, that this does not mean a smaller group needs to appear within one of the larger ones, only that a larger group cannot appear within a smaller group. Using these new hierarchical elements, the elements in the above coding example can be rewritten as: By creating these new hierarchical divisional elements, we not only enable a greater level of specificity in representing Cohn’s virtual groupings via elements, we further infer the hierarchical nature of these groupings. According to the proposed schema customization, cannot contain , and cannot contain either or , as the customization specifies these elements exist in a hierarchic structure. As such, the customized TEI/CBML markup understands that the virtual element created by the first is the largest group, followed by the virtual group, and last, the group. Despite their respective advantages and disadvantages, the above encodings all reflect one interpretation of Cohn’s theory regarding panel groupings. For example, readers may notice that, though Panel 1 is excluded from the two-panel Same Character group, it does not form its own single-panel Same Character group. Cohn’s visualizations for his hierarchic groupings—upon which Fig. 5 is modeled—do not make clear whether this first panel in the linear series that depicts the raven will constitute its own single-panel Raven group alongside the two-panel Poe group or will stand alone as a third wheel, so to speak, next to the Poe panels within its parent Shared Environment group (cf. Cohn 2010, figure 11). In contrast to the latter, which has already been represented in markup above, the former interpretation would be represented using Walsh’s
element as:
  
      
 
  
     
 
          
 
These structures can also be represented using the TEI element and our customized , , and elements as:
           
The issue with this interpretation of Cohn’s theory is one of logic, i.e. can a group of one exist? Specifically to the present context, can Panel 4 constitute a Shared Environment group if it does not have any panels with which it shares an environment? The present essay does not attempt to answer this question but only raises the issue to demonstrate how different markups of a comics document reflect not only diverging theories of comics but also divergent understandings of those theories. Much like the previous encodings mentioned in this section, the above two markup examples represent Cohn’s hierarchic groupings. The above two examples, however, differ from the previous encoding examples in the present section of this essay by interpreting Cohn’s hierarchic structure as consisting of panel groups formed from smaller groups—e.g. the Time Shift group is formed from two or more Shared Environment groups, and that these groups principally differ in their depicted times. Cohn never addresses whether a single panel lying outside a hierarchical group is interpreted as a one-panel group in itself or is to be approached as a mere outlier to a panel collective from which it is excluded. Much like the previous markups for Fig. 5 examined earlier, the above two examples reflect one interpretation of Cohn’s theory. Readers will notice that, when representing panel sequences or pages within Cohn’s framework, I have elected to use Walsh’s
element rather than the element previously employed for representing Groensteen’s arthrology. The decision to forego the element when representing Cohn’s theory stems from my own belief that, contra Groensteen, Cohn displays less concern than Groensteen with the geometrical dimensions of the comics page and the layout’s function as a visual meaning-making symbol (Cohn, 2013a; cf. Bateman et al., 2017). Rather, Cohn’s attention focuses primarily on how readers cognitively process panels in sequence. Indeed, Cohn suggests at times that the layout's geometrical proportions and spatiality are potentially irrelevant to reader processing, such as when Cohn writes that page layouts ‘do not appear to significantly impact reading comprehension’, which leads to his driving research question: ‘[H]ow do people know how to navigate through page layouts?’ (Cohn, 2013a). In answering this question, Cohn develops his idea of assemblage, which seeks ‘to build successive units of structure based on distance and coherence of composite shapes in as smooth a reading path as possible’ (Cohn 2013a; emphasis in original). In response to my discarding of the element when representing Cohn’s theory in markup, some readers may point to this last passage as testifying to the importance Cohn places upon the spatial relationship between panels. After all, he suggests that features such as the position of panels, like inset panels, can impact reading order across the page. For some, this nod to spatiality may warrant the use of the element. In the end, the decision to represent a page in markup according to Cohn’s framework by utilizing the element or a mere
depends upon how one interprets Cohn’s approach to the spatiality of the comics page. As shown throughout this paper, how one encodes a comics document depends not only on the adopted critical theoretical framework but also how an encoder or group of encoders interprets that framework. 7 Conclusion As may be observed in both analyses of Cohn and Groensteen’s theories, markup offers both a metalanguage for describing a given document or document type as well as a metacriticism for representing and analyzing critical approaches to those documents. Both Groensteen and Cohn advance well-developed theoretical frameworks for understanding the comics medium, and this essay by no means aims to debunk or argue against either theories. Rather the application of markup language in representing their respective frameworks aims to demonstrate how both, no matter how meticulous the original theoretician may be, require some level of interpretation by the reader—these different interpretations represented in different approaches to implementing markup language. Encoding a comics document not only requires that the encoder adopts a specific critical framework to comics, but that the encoder also adopts a specific understanding of that framework; the markup not only reflects one’s critical approach to comics but also one’s understanding of that critical approach. As such, markup language operates as a form of metacriticism for interpreting, representing, and modifying any given critical or theoretical framework. As both a metalanguage for describing documents and a meta-criticism for understanding critical frameworks, markup languages like TEI and CBML provide, echoing the words of Jerome McGann that opened this essay, ‘a dynamic engagement with [comics] and not a program aimed at discovering the objectively constitutive features of what a [comic] “is”’ (McGann, 2002). By encoding comics documents with TEI and CBML, one not only produces searchable, encoded versions of comics for digital collections but also actively engages with the massive variety of extant comics documents and numerous critical frameworks by which these documents can be approached. In other words, markup languages help critics reflect on the function and nature of comics, individually and collectively, while examining their own theoretical and critical frameworks. In this essay, I have largely examined Walsh’s methods for encoding panels and panel groups in comics, suggesting ways his practices may be reformulated to fit alternative theories of comics. Similar analyses may be applied to his methods for encoding text balloons or narrative captions. Moreover, there presently exists no systemized method for encoding comics' variety of graphic signs, e.g. heart eyes or dizzy stars. All of this is to say: there remain many aspects regarding the application of markup language to comics that require further research. In building upon Walsh’s foundational work, I hope this essay instigates further inquiries and advancements in developing a markup language for comics. References Bateman J. , Veloso F., Wildfeuer J., Cheung F., Guo N. ( 2017 ). An open multilevel classification scheme for the visual layout of comics and graphic novels: Motivation and design . Digital Scholarship in the Humanities , 32 ( 3 ): 476 – 510 . OpenURL Placeholder Text WorldCat Chute H. ( 2008 ). Comics as literature? Reading graphic narrative . PMLA , 123 ( 2 ): 452 – 65 . Google Scholar Crossref Search ADS WorldCat Cohn N. ( 2010 ). The limits of time and transitions: Challenges to theories of sequential image comprehension . Studies in Comics , 1 ( 1 ): 127 – 46 . Google Scholar Crossref Search ADS WorldCat Cohn N. ( 2013 a). Navigating comics: An empirical and theoretical approach to strategies of reading comic page layouts . Frontiers in Psychology , 4 ( 186 ): 1 – 15 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Cohn N. ( 2013 b). The Visual Language of Comics: Introduction to the Structure and Cognition of Sequential Images . London : Bloomsbury . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Cohn N. ( 2005 ). Un-defining ‘comics’: Separating the cultural from the structural in comics . International Journal of Comic Art , 7 ( 2 ): 236 – 48 . OpenURL Placeholder Text WorldCat Davies P. F. ( 2019 ). Comics as Communication: A Functional Approach . Cham : Palgrave Macmillan . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Drucker J. ( 2009 ). SpecLab: Digital Aesthetics and Projects in Speculative Computing . Chicago : University of Chicago Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Eisner W. ( 1985 ). Comics and Sequential Art . New York, NY : Poorhouse Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Flanders J. ( 2005 ). Digital Humanities and the Politics of Scholarly Work. Ph.D. thesis, Brown University. ProQuest Dissertations and Theses. https://search-proquest-com.ezproxy.neu.edu/docview/305028561?pq-origsite=primo (accessed 25 March 2020). Groensteen T. ( 2007 ). The System of Comics . Beaty Bart, Nguyen Nick (trs). Jackson : University Press of Mississippi . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Groensteen T. ( 2013 ). Comics and Narration. Miller Ann (tr.). Jackson : University Press of Mississippi . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Hayman G. , Pratt H. J. ( 2005 ). What are comics? In Goldblatt D., Brown L. (eds), A Reader in Philosophy of the Arts . Upper Saddle River, NJ: Pearson Education Inc , pp. 419 – 24 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lefevre P. ( 2010 ). Intertwining verbal and visual elements in printed narratives for adults . Studies in Comics , 1 ( 1 ): 35 – 52 . Google Scholar Crossref Search ADS WorldCat Manovich L. ( 2012 ). How to compare one million images? In Barry D. M. (ed.), Understanding Digital Humanities . London: Palgrave Macmillan , pp. 249 – 78 . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC McCloud S. ( 1993 ). Understanding Comics . New York, NY : Kitchen Sink Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC McGann J. ( 2002 ). Dialogue and interpretation at the interface of man and machine: reflections on textuality and a proposal for an experiment in machine reading . Computers and the Humanities , 26 : 95 – 107 . Google Scholar Crossref Search ADS WorldCat McGann J. ( 2001 ). Radiant Textuality: Literature After the World Wide Web . Palgrave Macmillan . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Mikkonen K. ( 2017 ). The Narratology of Comic Art . New York, NY : Routledge . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Miodrag H. ( 2013 ). Comics and Language: Reimagining Critical Discourse on the Form . Jackson : University Press of Mississippi . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Mitchell W. J. T. ( 2014 ). Comics as media: Afterword . Critical Inquiry , 40 ( 3 ): 255 – 65 . Google Scholar Crossref Search ADS WorldCat Peeters B. ( 2007 ). Four conceptions of the page . ImageTexT: Interdisciplinary Comics Studies , 3 ( 3 ). Jesse Cohn (tr.). http://www.english.ufl.edu/imagetext/archives/v3_3/peeters/ (accessed 25 March 2020). OpenURL Placeholder Text WorldCat Tanaka V. ( 2008 ). Metronome . New York, NY : NBM Publishing . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC TEI. ( 2020 ). TEI P5: Guidelines for Electronic Text Encoding and Interchange. TEI Consortium. https://www.tei-c.org/release/doc/tei-p5-doc/en/html/index.html (accessed 25 March 2020). Walsh J. ( 2012 ). Comic book markup language: an introduction and rationale . Digital Humanities Quarterly , 6 ( 1 ). http://www.digitalhumanities.org/dhq/vol/6/1/000117/000117.html (accessed 25 March 2020). OpenURL Placeholder Text WorldCat Witek J. ( 2009 ). The arrow and the grid. In Heer J., Worchester K. (eds), A Comics Studies Reader . Jackson : University Press of Mississippi , pp. 149 – 56 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Footnotes 1 In documentation on XML markup languages (of which TEI and CBML may be understood as species types), element names are wrapped in angle brackets (< >) and attribute names are preceded with the @ sign. The present essay follows these conventions throughout. 2 The essay here follows Groensteen’s differentiation between a panel series and sequence, the former being ‘a succession of continuous or discontinuous images linked by a system of iconic, plastic, or semantic correspondences… A sequence is a succession of images where the syntagmic linking is determined by a narrative project’ (Groensteen, 2007) . 3 For convenience, the figure’s coordinate space is here defined according to the pixel dimensions of the original digital image file, with each pixel being treated as one unit across the coordinate plane. Thus, the original file is 736 × 1,142 pixels. 4 Zero values are here used because the given panel runs to the edge of the page. 5 I imagine most readers familiar with TEI would adopt this encoding for the simple reason that, unless the two pages share layout and content, they are, quite simply, not the same. But as I have sought to demonstrate, this decision is not only determined by one’s knowledge of TEI or Groensteen’s critical theoretical framework but also how one interprets Groensteen’s framework. 6 Available for download at: http://dcl.ils.indiana.edu/cbml/ (accessed 23 April 2020). © The Author(s) 2020. Published by Oxford University Press on behalf of EADH. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) TI - On the use of XML markup language in comics criticism JF - Digital Scholarship in the Humanities DO - 10.1093/llc/fqaa028 DA - 2004-04-01 UR - https://www.deepdyve.com/lp/oxford-university-press/on-the-use-of-xml-markup-language-in-comics-criticism-DaJPUMWq0B DP - DeepDyve ER -