Between consciously crafted and the vastness of context: collateral paratextuality and its implications for translation studies

ABSTRACT This article presents a refined theory of paratextuality and bridges the gap between Kathryn Batchelor’s understanding of paratexts as a “consciously crafted threshold” and the risk of paratext’s “collapse into the vastness of ‘the context’” through a new form of paratextual relationship: collateral paratextuality. Doing so facilitates investigations into how audiences engage with complex constellations of paratextual materials in digital spaces, rather than analyzing paratexts in isolation. Such an approach is particularly relevant within translation studies given the proliferation of media content across linguistic and cultural borders within our increasingly networked and algorithmically mediated digital world. To demonstrate the utility of collateral paratextuality, this article concludes with a case study of the teaser trailer for Netflix’s Dark as hosted on YouTube, where the coalescence of content from different paratextual creators, YouTube’s mediation thereof, and the personalized nature of the platform both justifies and necessitates a collateral paratextual approach.


Introduction
Since the term's inception by Gérard Genette (1982, 1987), the concept of "paratext" has been broadened and expanded well beyond its original, French, literary context.For instance, scholars working in fields such as translation studies (Hong 2019), media and communication studies (Stanitzek 2005;Gray 2010) and film studies (Caldwell 2011) have developed the concept beyond Genette's focus on the elements found within and surrounding a book that are "more or less legitimated by the author" (1997a, 2) to include a kaleidoscope of materials including by-lines, trailers and memorabilia.However, this opening out of paratextuality as a theoretical concept has also necessitated a movement away from Genette's focus on authorial intent, with some scholars subsequently cautioning that taking these steps too far risks the collapse of paratextuality into "the vastness of 'the context'" (Rockenberger 2014, 267).How then can we approach and understand the paratextual function that an increasing number of materials can fulfill and the ways in which they coalesce within a given textual space, without undermining the value of the term through everything becoming paratext, particularly in contemporary, hyper-connected, digital contexts of ever-proliferating content?
Within this article, I propose the concept of collateral paratextuality as a solution to this challenge, a solution that continues to uphold the academic value of paratextuality as a concept whilst providing a mechanism through which to analyse complex constellations of paratextual materials found in digital spaces.To do so, I first present a refined definition of paratextuality that draws on work in translation studies, most notably Kathryn Batchelor's notion of "consciously crafted" paratextuality (2018), to refocus scholarly attention on a relational understanding of paratextuality that transcends material and disciplinary boundaries such as book/film.I demonstrate the limitations of conscious crafting, particularly within digital contexts, and establish how collateral paratextuality frees paratextuality therefrom.Finally, the utility of my approach both in terms of transdisciplinary understandings of paratextuality and research within translation studies is demonstrated through a case study of an English-subtitled teaser trailer for the German-language television series Dark as hosted on YouTube.
Defining paratextuality and the boundaries of the "consciously crafted" In this article, I understand paratextuality as a functional relationship between texts in which an encounter with one can frame an encounter with another within a given context.In line with multimodal understandings of meaning making, I argue that this function, and so the establishment of a paratextual relationship, can be fulfilled in an act of either creation or interpretation (Kress 2020, 32) and is understood as both "dynamic" and "culturally specific" (Kaindl 2020, 51).Thus, a paratextual relationship can be formed and identified from the perspective of both a producer and a receiver.Furthermore, a paratextual relationship is understood as one potential relationship that can exist between texts and as a relationship that can exist between any two texts.As such, the ontological question of whether X is a paratext of Y becomes redundant, thereby avoiding the need for taxonomic cataloguing of specific materials that privilege the materials found within the researcher's corpus and exclude any that could similarly form a paratextual relationships in other contexts.Rather, the primary question when delineating a paratextual corpus becomes "can a paratextual relationship exist between X and Y?", with the emphasis placed on the existence of this relationship within a given context.Notably, the use of the term "text" within this definition refers to "any multimodally composed meaningful whole" (Adami and Ramos Pinto 2020, 73), thereby comprising both the media texts and products at the heart of media studies, such as films and televisions series, and the promotional materials and audience responses found within what Jonathan Gray refers to as "off-screen studies" (2010, 7), such as reviews and online fan communities.As will be discussed in the next section, paratextual relationships can exist between all of these forms of text.
The definition of paratextuality put forward here continues recent movement towards the prioritization of a paratext's function over, for instance, its authorship (e.g.Batchelor 2018), whilst also drawing on the original context within which Genette's paratextuality was introduced and returning to the concept's relational roots.Genette first introduced paratextuality in Palimpsestes: La littérature au second degré (1982, translated into English as Palimpsests: Literature in the Second Degree [1997b]), as one of five types of transtextual relationship: intertextuality, paratexts, metatextuality, hypertextuality and architextuality.For Genette, these transtextual relationships constitute "all that sets the text in a relationship, whether obvious or concealed, with other texts" (1997b, 1).What is notable here is that Genette introduces intertextuality, metatextuality, architextuality, hypertextuality and the overarching transtextuality as abstract nouns that describe a quality.For instance, an intertextual relationship becomes intertextuality.In this way, Genette positions these forms of textual transcendence as potential relationships that can exist between any texts.In contrast, Genette defines paratext as the relationship between a text and a specific set of subordinate texts: "its paratext" (1997b, 3; "son paratexte" in the original French, 1982).This instantly raises a significant, ontological question for any scholar wishing to use paratext theory in their own work: what is the paratext?For Genette's other transtextual relationships, such a structuralist and hierarchical delineation is unwarranted.However, in Palimpsests Genette frames the paratext as a specific, subservient text type to be catalogued, rather than taking an abstract relational approach as with the other forms of textual transcendence.
Genette expands on this approach in his subsequent book, Seuils (1987, translated into English as Paratexts: Thresholds of Interpretation [1997a]), which focuses solely on paratextuality.Where his definition of paratext in Palimpsests allows for the inclusion of materials whether they are "official or not" (1997b, 3), however, Genette begins Paratexts with a refined definition that now includes the requirement that paratexts are "more or less legitimated by the author" (1997a, 2).With the boundaries of paratext more firmly defined in line with authorial legitimation, Genette then uses the book to inductively study such materials as found in the literary text manifest as a codex and taxonomically "follow the order in which one usually meets the messages this study explores: the external presentation of a bookname of author, title, and the restjust as it is offered to a docile reader" (1997a, 3).In doing so, Genette cements the original notion of paratextuality within a specific set of codex-based texts and practices, with the term "paratexts" referring primarily to these materials, rather than to the relationship they share with the literary text.
Within digital and media text contexts, however, Genette's inductive book-based approach to defining paratextuality is limited for several reasons.Firstly, the inductive approach taken throughout Paratexts results in a theory of paratextuality that is inherently embedded in the form of the codex and the practices of book production that were prevalent within Genette's context, thus requiring considerable adaptation for digital and audio-visual media.Subsequent attempts to apply Genette's paratext theory to new contexts have, therefore, resulted in further inductive studies with limited scope for wider application.For example, Dorothee Birke and Birte Christ (2013) find that the modes of textual production and reception available for digitized literature are so different from those studied by Genette that they must inductively catalogue and retheorize paratextuality for e-books and DVDs (66).Similarly, the electronic literature and e-reading devices studied by Ellen McCracken (2013) are so different from a physical book that she must again take an inductive approach with these new materials.Away from the study of literature, Annika Rockenberger (2014) also takes an inductive, auto-ethnographic approach to understanding the various paratextual materials that accompany video game texts, while Thomas Simonsen (2014) catalogues the paratexts used to frame and present YouTube videos.In all such cases, applications of a Genettian approach to paratextuality outside a literary context requires that we replicate his inductive approach to create new paratextual taxonomies that continue to primarily be relevant within individual research contexts.Of course, it would be unreasonable to criticize Genette's inability to predict the invention and popularity of new forms of media since the 1980s.Yet, if we return paratextuality to the relational roots of Genette's textual transcendence, such limited foci become unwarranted as this paratextual relationship can exist between any texts in any modality.
As such, I argue that paratextuality must be understood as a relationship, in line with Genette's original conceptualization of textual transcendence, rather than presenting a new definition of the term "paratext".This latter approach was taken by Kathryn Batchelor in her functional redefinition of paratext for translation studies contexts, in which she sought to similarly avoid taxonomic delineations of where the paratext starts and ends by referring to "a paratext" (2018,142).The limitation of this approach is that despite the use of an indefinite article, which opens the possibility of any material serving a paratextual function and so being "a paratext", such a definition continues to point analysts to the binary question of "is this a paratext?"and thereby facilitates the continuing creation of inductive or taxonomic studies with specific relevance and limited scope for wider application.Consequently, by returning to paratextuality and a relational understanding of the term, I restore the theoretical concept to the potential of its origins and free the concept from any one context or text type.

Constellations in the paratextual space
Given that the present article understands paratextuality as a relationship that can exist between any two texts, rather than a specific set of texts, it must be noted that this relationship is not understood as hierarchical.Rather, paratextual relationships set texts in rhizomatic (Deleuze and Guattari 1987) constellation with one another and may move audiences towards, around, or away from any given text.In this way, all texts are understood as unique and distinguishable texts in their own right that can be connected to other texts through the establishment of paratextual relationships.Thus, where I identify constellations of paratextual relationships, the various texts within are positioned as "co-texts" (Adami and Ramos Pinto 2020) that co-occur and co-construct meaning through their relationships both with the media text and each other.When seeking to map and understand these constellations and the trajectories of audiences around media texts, then, the analyst must position these texts within a paratextual space to ensure that the rhizomatic nature of a constellation of paratextual relationships is retained and remains non-hierarchical.Depending on the scope of a given project, this paratextual space may span a singular web-page or physical artefact, or may comprise a vast quantity of paratextual materials spanning various formats and modalities.
This approach has several benefits for paratextual analysis.Firstly, within his original conceptualization of paratextuality Genette takes a production-side approach by subordinating the paratext to the literary text and positioning the functional significance of paratextuality in terms of the purpose that paratextual creators want their texts to serve: to draw readers towards the more significant literary text.Yet, as noted by Gray (2010) and as implied by Genette's distinction between the ways in which a literary text is offered "to its readers, and more generally, to the public" (Genette 1997a, 1, my emphasis), audiences do not simply encounter paratexts for the media texts they choose to read or watch; they also encounter paratexts for media texts that they will not engage with.Consequently, audiences encounter far more paratexts than they do media texts and so, from a reader-based perspective, paratexts can be as, if not more, significant points of contact between an audience and a media text.Thus, situating all forms of text, including the media text and marketing materials for said media text, as co-texts within the same paratextual constellation, rather than within a textual hierarchy, acknowledges the societal role that paratexts play and their relevance as texts in their own right within the present theoretical framework.
The second benefit to this approach stems from the fact that the paratextual relationships that construct these constellations are not tethered to particular materialities or spatial locations.As such, the present approach acknowledges that texts spanning different media and modalities, as well as the digital/non-digital divide, can form paratextual relationships with one another, rather than viewing a particular paratext as only pointing towards a specific version or material form of a media text.Consequently, the constellation approach to paratextuality defies binary distinctions such as Genette's "peritext" and "epitext" (1997a), which refer to materials found appended to the text and those that circulate independently.The problem with such distinctions lies in their privileging of a particular material form, which becomes untenable in digital spaces where hypertextual links, embedded content and an endless "sense of flow" (Brookey and Gray 2017) blur the lines between materials that circulate independently from the media text and those appended to it, as well as between all such materials and the media text itself.Instead, all such materials are understood as sharing a paratextual relationship with the media text whilst those that are found within the media text as product, be it as a physical DVD or a film made available through a digital streaming service, also serve as co-texts in that manifestation of the textual product.Thus, paratextual and co-textual relationships are neither binaries nor are they mutually exclusive.Furthermore, binarism in the spatial definition of paratextual spaces fails to acknowledge the complex trajectories that different receivers can take towards a media text.Indeed, as noted by McCracken (2013), different paratexts can move a reader both centripetally towards a given version of the text, or centrifugally away from the text entirely.In this way, viewing paratexts as constellations acknowledges the complex routes that receivers can take towards media texts and prevents the two-dimensional spatial binarism, and subsequent privileging of particular paratexts, that can result from Genettian approaches.
Finally, by conceptualizing paratextuality in the form of a rhizomatic constellation, the theoretical framework presented here understands paratextual relationships as fluid and dynamic, meaning that the existence of a paratextual relationship within a given historical, social, linguistic or cultural context does not result in a permanent state of paratextuality or preclude paratextual relationships from evolving or evanescing.As such, the question of whether a given paratextual relationship is substantial or relevant within a given context depends on the research questions asked, the scope of the analysis, and the focus of the study in question (as also noted by Batchelor 2018, 144).Consequently, all studies that use paratextuality as a tool to delineate a corpus of study cannot simply state that they will analyse the "paratexts" of a media text as imprecise shorthand for a particular constellation of materials.This inexactness is based upon the assumption that paratextual relationships take a fixed form that will be communicable outside of the researcher's own context and fails to acknowledge that as new technology develops outside of the researcher's foresight, so too will the potential forms of paratextuality.Rather, our definition of paratextuality must be as malleable as possible and so cannot limit its scope to paratextual roads well-travelled or to the paratextual relationship du jour.

Para-textual hybridity
By seeing paratextuality as one potential relationship that can exist between any two texts, the theoretical framework put forward here understands all texts as both works in their own right and as (potential) framing devices for other texts.As such, a functional, paratextual relationship is one of many potential relationships that can exist between texts.Indeed, it is through this duality that paratextuality gains a hybridity in which a previously hierarchical relationship can extend across multiple rewritings of a text into the rhizomatic constellations of paratexts introduced in the previous section.In this way, new versions or "rewritings" of a text (Lefevere 1992), such as a translation or adaptation, can simultaneously exist as texts embedded in constellations of paratexts and as paratexts within a constellation for another text.What distinguishes between these states is the creation or interpretation of this meaning by a paratextual creator or receiver, whose trajectory through a paratextual constellation can either result in the creation of new paratextual relationships or leave the potential for paratextuality between two texts unfulfilled.
While some have criticized understanding translated texts as paratexts as it subsequently positions translation as "derivative", such as Tahir Gürçağlar ( 2002), such arguments become redundant when we explicitly acknowledge paratextual hybridity because textuality and paratextuality are not mutually exclusive (see also Batchelor 2018, 28-30).Nevertheless, if rewritings such as a translation or filmic adaptation are understood as both texts in their own right and as paratexts to their source text, a temptation to formalize when a rewriting is or is not considered a paratext may arise.As discussed in the previous section, however, the issue with such lines of inquiry is that they conceive paratextuality as an ontological and privileged category rather than as a relationship that can exist between any texts.Within the present framework, then, rewritings such as translations or adaptations form a paratextual relationship with their source text when an encounter with the target text can frame a subsequent encounter with the source within a given contextor vice versa.Should this subsequent encounter never take place, then this potential paratextuality goes unfulfilled in this context.Notably, the definition of "encounter" used here remains open and includes any engagement that a receiver can have with a media text.Be it a glance at a promotional poster; reading a review online; or watching an entire film or television series; any form, length or level of encounter with a text can result in the formation of paratextual relationshipsparticularly when we consider that readers encounter far more paratexts than texts in their lifetime (Gray 2010), as noted above.

Paratextuality and authorship
Paratextual relationships can exist between two texts regardless of their authorship or whether they can be authorized by a figure involved in the production of the media text.In taking this approach, this framework facilitates deeper investigations into the various individuals involved in the creation of production-side paratexts, which refers to paratextual relationships created during the production of a media text, whilst also acknowledging the proliferation and increasing relevance of paratextual relationships that are crafted outside of textual production processes, such as those created by receivers.
By removing any requirement for authorship in my definition of paratextuality, I continue the transdisciplinary trend to move away from Genette's strict insistence that paratexts are "always the conveyor of a commentary that is authorial or more or less legitimated by the author" (1997a, 2).In monolingual and codex-focussed research contexts, such as publishing studies, this authorial responsibility remains a key element in conceptualizations of paratextualityeven in digital contexts (see, for instance, Murray 2018, 168-180).However, where disciplines such as translation studies, or media studies have a more complex conceptualization of authorship but have adopted at least the term "paratext" from Genette, the notion of authorial responsibility and legitimacy has begun to slip away.In translation studies, for instance, suggestions such as Sharon Deane-Cox's "translatorial" paratext have expanded Genette's typology to include materials specifically created by translators within the paratextual paradigm (2014, 29), while others define the term purely in pragmatic, functional terms and use quotes from Genette that omit passages specifically referring to auctorial authority (see Batchelor 2018, 27-28).
Movements away from Genette's authorship model can also be traced in film studies and media studies.Distinctions have been made, for instance, between industry created (Gray 2010, 143), and so "official" (Mittell 2015, 262;Boni 2016, 213), or viewer/audience created (Gray 2010, 143), and so "unofficial" or "grassroots" (Mittell 2015, 262;Boni 2016, 213) paratexts.Such differentiation reflects a distinction between "top-down" and "bottom-up" approaches to the study of paratextual creation (Caldwell 2011) that allows researchers to position the user-generated content of digital, Web 2.0 platforms alongside the paratextual materials created during textual production processes.Outside of the producer/receiver binary, scholars such as Sérgio Tavares (2015) have moved into digital paratextual creation processes through the concept of "network paratext" for algorithmically generated content, whilst Gray (2010) has actively removed authorship as a paratextual criterion altogether.Within these alternative approaches, the focus has moved towards materials that influence the ways in which audiences receive and encounter texts that are created by other audience members and so can be seen as receiver oriented.Yet, whilst this expansion of paratextuality allows us to incorporate new forms of paratexts and paratextual relationships within our definitions, critics have noted that such a move risks undermining the significance of paratextuality as a theoretical concept by allowing everything to become paratext (Rockenberger 2014, 267;Batchelor 2018, 143).
An attempt to find a middle ground can be found in Batchelor's Translation and Paratexts, where she suggests the following definition: "A paratext is a consciously crafted threshold for a text which has the potential to influence the way(s) in which the text is received" (2018,142).Here, Batchelor has removed all explicit references to the author and thereby freed the notion of paratextuality from the control of an authoritative figure to allow research that "takes a producer-based approach" as well as a "receiver-based perspective" (2018,143).In doing so, Batchelor has also facilitated the inclusion of user-/audience-/reader-created materials within the paratextual space and presented a framework that, as demonstrated in her Walter Presents case study in the same monograph (2018,, can situate a text within the digital paratexts that proliferate around it.Yet, the problematic trace of authorship remains in Batchelor's definition. While the main caveat placed within Batchelor's redefinition of paratextuality, the notion of "conscious crafting", places the author as a more distant figure among a myriad of paratextual creators, it does not remove all authority figures from the paratextual space.Where I argue that a paratextual relationship can exist between any two texts, Batchelor's caveat "places broader context as well as happenstance (for example, the individual circumstances through which a particular reader comes to a text) outside of the scope of the paratext" (2018,142).Batchelor is keen to note that this is not to "ignore the fact that contextual and incidental factors can have a bearing on the way a text is received" (2018, 143) but rather seeks to prevent what Rockenberger describes as the "paratext's collapse into the vastness of 'the context'" (2014,267).While the openness of this definition allows the answer to the question "consciously crafted by whom?" to be answered in the broadest sense, and thereby allow receivers to become paratextual creators, the issue becomes how we identify this crafting and how we understand materials that were not "consciously crafted" as thresholds for a text but may still function as such.Indeed, Batchelor's definition runs counter to the multimodal understanding of meaning making that underpins this theoretical framework in which paratextual relationships can be made in both the act of creation (or in Batchelor's terms, crafting) and interpretation.Rather than requiring paratextual analysts to identify the presence of an authorial figure, then, Batchelor's definition shifts this requirement on to identifying the conditions of a paratext's creation.
In a world where both digital and non-digital texts proliferate outwards into an insurmountable quantity of content, however, such identification is not always practical or possibleparticularly when dealing with digital materials that are generated algorithmically (in part or in full) or materials that are created to paratextually frame one version of a text but then also form paratextual relationships with another version or rewriting.As such, I suggest the identification and study of a new form of paratextual relationship: collateral paratextuality.

Collateral paratextuality
Now that the definition of paratextuality used here has been clarified, I present a new form of paratextual relationship that falls outside the notion of Batchelor's "consciously crafted" paratextual relationship (2018, 142) without collapsing paratextuality into "the vastness of 'the context'" (Rockenberger 2014, 267).I refer to this form of paratextual relationship as collateral paratextuality, which I define as a paratextual relationship that is formed inadvertently, in parallel or in addition to another without the conscious intervention of a creator.I do not situate these two forms of paratextuality within a hierarchical structure and so neither a "consciously crafted" nor a collateral paratextual relationship is subordinate to the other.Rather, collateral paratextuality is here introduced to explicitly acknowledge that while some texts may be created and published with the intention that they form a paratextual relationship with a given media text, this is not always the case.In some cases, paratextual relationships form without the conscious intervention of an authority as the consequence of an individual creator's or interpreter's engagement with a text.Yet, by placing these two forms of paratextuality in constellation alongside one another, I emphasize that for the receivers who traverse and construct paratextual constellations, any form of paratextual relationship can impart as much influence over their engagement with the media text as another.
There are several benefits to including collateral paratextual relationships within our understanding of paratextuality.Firstly, the introduction of collateral framings emphasizes that paratextual relationships can be formed when a text is engaged with and interpreted, rather than simply created.Doing so allows us to include wider textual materials that can influence the way receivers approach media texts within our understanding of paratextuality without any requirement for conscious craftinga move that is particularly necessary in digital spaces where paratexts are not typically viewed in isolation.Thus, by acknowledging that the materials that sit alongside consciously crafted paratexts as co-texts on a given platform can also influence the way audiences approach subsequent textual encounters, we require the concept of collateral paratextuality to understand these relationships.Furthermore, understanding paratextuality as a relationship between texts prevents the "paratext's collapse into the vastness of 'the context'" (Rockenberger 2014, 267) by maintaining our focus on the textual, even when expanding out to collateral paratextuality.For instance, a viewer's prior knowledge of a director or actor, such as their gender, sexuality, race or background, may not stem from paratextual materials that were created in relation to a specific media text, but may instead come from their other work or public appearances.Within Batchelor's definition, such details must be considered "broader context" or "happenstance" (2018,143), rather than paratextual, unless they are found in materials that fall into the category of "consciously crafted threshold".However, broader textual manifestations of their work can frame the way a receiver encounters a subsequent text and so fulfill a paratextual function even when this paratextual relationship is not consciously crafted.Thus, collateral paratextuality presents a lens through which to examine how textual manifestations of "broader context" can serve a paratextual function.
Notably, Genette also sought to include "a fact whose existence alone, if known to the public, provides some commentary on the text and influences how the text is received" within his definition of paratextuality, referring to it as "factual paratext" (1997a, 7).However, while this would seemingly cover the hypothetical given above, it can easily run counter to Genette's requirement for paratexts to be "more or less legitimated by the author" (1997a, 2).Furthermore, this article understands paratextuality as a relationship between texts, yet Genette's factual paratext includes materials that "may or may not be brought to the public's attention by a mention that, itself, belongs to the textual paratext" (1997a, 8).As such, Genette's conceptualization of factual paratextuality not only contradicts his own definition of paratext, but also facilitates what Rockenberger describes as "the paratext's collapse into the vastness of 'the context'" (2014, 267) by including facts that exist well beyond the textual.By turning to collateral paratextuality within the present theoretical framework, then, we can investigate how textual manifestations of broader contextual information can frame a receiver's subsequent encounter with a media text.In doing so, texts that were not consciously crafted as paratextual materials, but whose content can influence the ways in which receivers approach and understand a media text, are brought firmly into view within discussions of this paratextual function.
Secondly, the concept of collateral paratextuality is beneficial in digital spaces, where the intervention and mediation of algorithmic and user-generated data is omnipresent.In digital spaces, such algorithms can be used to generate personalized recommendations for users and provide hyperlinks to other texts, thereby creating complex trajectories for a user to take that are created solely by machine-programmed sets of rules and data.An example of this is YouTube's "deep neural networks for recommendations", which leverages a "user's YouTube activity history" and collaborative filtering of other user's viewing histories to generate video recommendations for anyone visiting the site (Covington, Adams, and Sargin 2016).In such cases, the data used within these algorithms is generated by cookies, which are small files containing details about website usage that are saved to a user's device when they connect to said website.As this cookie data is userspecific, the paratextual relationships formed by such algorithms present users with bespoke paratextual constellations to traverse.While Batchelor uses scholarship from digital studies to argue that "the medium through which a text is discovered", including search engines and hyperlinks, can constitute a form of paratext (2018,155), I contend that this is incompatible with her definition of paratextually because the individualized nature of these recommendations means they explicitly fall within "the individual circumstances through which a particular reader comes to a text" (143). 1 The individualistic nature of these algorithms can also result in a single user engaging with different digital paratextual constellations depending on the device they use, as individual devices store and report their own specific cookie data.Given that no human agent can take responsibility for the creation of these individualized paratextual relationships, only the creation of the set of rules constituting the algorithm's code, I maintain that these relationships lie outside of the scope of Batchelor's "consciously crafted" thresholds (2018,142).Yet their prevalence in digital spaces requires that we take them into account when questioning how media texts are presented and circulated online (Brookey and Gray 2017).Thus, within the notion of collateral paratextuality, we can include and analyse such paratextual relationships.

Demonstrating the utility of collateral paratextuality: the YouTube teaser trailer for Dark
To further demonstrate the ways in which scholars within translation studies and media studies can approach and analyse collateral paratextuality, I will now present a brief case study of the teaser trailer for Dark as hosted on the US Netflix YouTube account.I suggest that there are three ways in which we can understand materials found on the YouTube video page as forming collateral paratextual relationships with the media text. 2 In this case, the entire webpage as it is found at https://www.youtube.com/watch?v=ESEUoa-mz2c is understood as a paratextual space, with the various materials found within serving as co-texts in the construction of this space and forming paratextual relationships with the audio-visual text of the television program Dark.For context, Dark is a German-language, science-fiction television series released on Netflix with German audio and translated subtitles.The teaser trailer discussed here was published on YouTube on 1 March 2017, with the first series then released on 1 December of the same year.At the time of writing, this trailer has amassed approximately 1.5 million views.The audio-visual elements of the trailer feature the original German-language audio with embedded English-language subtitles provided by Netflix themselves, whilst all other production-side textual elements, such as the title and video description, are provided in English.

Paratextual coalescence
The first way in which the trailer for Dark as hosted on YouTube can be understood through the lens of collateral paratextuality comes on a macro level through the coalescence of elements crafted by various creators for which no single entity takes authorial or editorial responsibility within the singular textual space available at https://www.youtube.com/watch?v=ESEUoa-mz2c.For instance, elements such as the audio-visual trailer, video title and video description are provided by Netflix as the video uploader, whilst the user comments found at the bottom of the page are written and posted by individual site users with no collective identity or institutional affiliation, and video recommendations are generated through the site's algorithms and individual user activity.Nevertheless, all of these elements are made available to site users at a singular location and come together to make meaning as the webpage, thereby serving as co-texts in the composition of this collateral paratextual space.While YouTube is responsible for the platform's existence and the form that pages within this platform take, thereby defining the type of content that can be featured therein, the site does not approve, moderate or curate the content found in individual videos or user comments thereon.Indeed, the fact that "many hours of video are uploaded per second" would make such moderation almost impossible (Covington, Adams, and Sargin 2016).As such, the coalescence of materials with varied authorship within a singular digital space for which no one entity can take responsibility, and the automated content found on these pages, means that the pages cannot be defined as "consciously crafted" thresholds.
This does not mean that such materials have not previously been the subject of paratextual inquiry.Indeed, Batchelor's own case study of subtitled television programs within the Walter Presents brand makes use of sources such as YouTube comments and viewing figures, as well as the brand's self-presentation on their own website (2018,.However, such studies, including Batchelor's, typically take this content in isolation and qualitatively discuss elements like user comments outside their position as co-texts within the entire webpage in which they are embedded and users encounter them.YouTube users do not read user comments on a video in isolation, they do so in the context of an entire webpage that also features consciously crafted production-side materials listed above, as well as materials such as algorithmically generated recommendations and video advertisements.Collateral paratextuality therefore allows us to view this entire webpage as a paratextual space, rather than extracting individual elements and analysing outside the context in which users encounter them.Furthermore, while YouTube as an organization may be responsible for the overall design and general layout of video pages on their website, no one individual or institution can take responsibility for all of the forms of content that coalesce within this constellation.Yet, in the real world they all come together to make meaning for site users who encounter this as one textual space.I would even go so far as to argue that the view counter discussed by Batchelor must also be considered collaterally, as no individual author can take responsibility for its content or what this number communicates paratextually to site users.Thus, both the constellation-based approach developed here and the concept of collateral paratextuality are required to properly situate individual paratextual materials within the broader contexts in which users encounter them and to investigate how all these elements co-construct paratextual meaning. Of course, YouTube is not the only digital platform where materials created by different paratextual creators coalesce within a singular textual space, with other digital and online platforms such as Amazon providing a similar space for this paratextual coalescence (cf.Freeth 2022, 200-240).Rather, the position of YouTube as the most popular online video hosting platform in the West, as well as the co-presence of production-and reception-side materials within YouTube's individual video pages, demonstrates that while the site owner may take responsibility for their platform's presence and continued availability within the digital sphere, the conflation of paratextual creators results in the creation of complex yet influential textual spaces for which no one creator can take complete authorial responsibility. 3What's more, these spaces remain in a constant state of mutability as new site users are able to add content at any time.As such, the structure of digital and online platforms such as YouTube presents a uniquely digital challenge to the boundary between productionand reception-side paratextual creation by conflating the two within a singular, collaterally created digital space.

Mediation of content as collateral paratextuality
More specifically in terms of the teaser trailer for Dark, this paratextual coalescence and YouTube's mediation thereof results in an interesting disparity between the two paratextual narratives put forward by the production-and reception-side materials.In the case of the former, this teaser was released nine months before the television series, thereby constituting what Gray (2010) would describe as an "entryway" paratext that was created to be consumed prior to the main televisual text.Notably, however, YouTube automatically sorts the user comments for the teaser trailer into an order that shows "top comments" first.This ranking system is calculated algorithmically by YouTube with a translucent aggregated system including various factors, such as when a comment was posted, the ratio of "likes" or "dislikes" a comment receives, and the number of replies by other users.In the case of the Dark teaser, this results in top comments created by viewers who became paratextual creators after encountering the text.For instance, the first comment by 'The Spook' asserts "This, ladies and gentleman [sic], is the most beautiful and mindblowing show ever made by humans" whilst 'The Social Trends' exclaims "Just finished Season 3. The best ending of a show.Came here to relive some amazing moments".In discussing their viewing experience of the program post-hoc within the same paratextual space as a teaser trailer designed as an entryway to create initial viewer interest, the coalescence of production-and reception-side paratextual materials creates a complex temporal web of paratextuality that, as a whole, cannot be understood as consciously crafted.In taking a collateral approach, however, we are still able to understand the, in Gray's terms (2010), "entry-way" production-side framing and the "in-media res" reception of (translated) user comments as a singular paratextual space.
What's more, the only other comment-sorting option offered by YouTube is "Newest first", which similarly displays comments that refer to Dark as a whole seriesrather than the first series for which this teaser was crafted.Writing just days before I conducted this analysis, for instance, 'Kristi Marie argues that "[T]he first season was interesting, but by season 3 it got confusing".As such, it is impossible to sort user comments by oldest first and so present visitors to this page with an experience that mirrors that of a user first viewing this teaser before the program launched in 2017.Instead, the teaser for series one on YouTube inadvertently forms collateral paratextual relationships with later series, a purpose for which it was not initially craftedparticularly in the case of series three, which was filmed in 2019, two years after the program's initial release.This, therefore, also demonstrates the value of understanding the paratextual space as a dynamic rhizomatic constellation of texts through which receivers can take infinitely varied trajectories both towards and away from a media text.

Personalized paratextuality
The final example of collateral paratextuality identified here comes from the use of algorithmically generated texts to create personalized paratextual spaces.Where YouTube's algorithmic intervention in the sorting of user comments mediates the paratextual space for all site users at a given point in time, the use of recommendation algorithms to present users with videos deemed to also be of interest is a uniquely personal process that poses difficult challenges to researchers.For instance, when loading the Dark teaser trailer on YouTube whilst logged into my own personal account (see Figure 1), I primarily receive recommendations for content based on my previous viewing history, such as music videos and Formula 1 coverage, alongside a limited selection of content relating to other sci-fi on film and television.Consequently, should any of my subsequent encounters with these recommended videos be framed by my viewing of the teaser for Dark, this paratextual relationship must be deemed collateral as this video was clearly not created as a framing device for an unrelated music videothough commonalities such as the artist and television show both being German may still exist, thereby allowing the shared Germanness of the two videos to create a paratextual link between the two for me as an individual.
However, as discussed by Covington, Adams, and Sargin (2016), YouTube's recommendation engine leverages not only an individual user's viewing history but also the viewing histories of other users who watch the same videosparticularly in cases where little-to-no specific user data exists.These recommendations can, therefore, also reflect the broader contexts in which particular YouTube videos were received by giving an indication of the interests of other viewers who watched this video.To see this in action we can visit the same URL for the Dark teaser without logging into a YouTube account and within a so-called "private browsing" mode, where individual user data is no longer stored.Here (see Figure 2), we see the recommendations for videos relating to other sci-fi media texts, such as Westworld and Dune, as well as real-world space exploration.This thereby indicates that YouTube visitors interested in the teaser trailer for Dark series one are also interested in English-language sci-fi, rather than other successful German-language Netflix series, such as the historical fantasy series Barbarians.If then, we imagine a user viewing the teaser trailer for Dark who has never seen the program but who has seen Westworld, seeing this recommendation creates a paratextual link between the two series that was not consciously crafted by the production teams of either but yet communicates the sci-fi nature of both programs.As such, this specific case study demonstrates the complex ways in which users traverse paratextual spaces and the significance of collateral paratextual relationships in understanding their trajectories therein.

Conclusions
The present article has sought to expand current understandings of paratextuality by incorporating what I define as collateral paratextuality.The identification and analysis of this form of paratextual relationship seeks to provide insights into the ways digital spaces construct meaning and forge paratextual relationships through the inclusion of complex constellations of materials for which no one individual or institution can take responsibility.Doing so facilitates a better understanding of how audiences approach and frame media texts outside of Batchelor's "consciously crafted" paratextual space by examining paratexts as they are engaged with by audiences, rather than in analytic isolation, and without slipping into "the vastness of 'the context'" (Rockenberger 2014, 267).As demonstrated by the brief case study of a teaser trailer for series one of Netflix's Dark as hosted on YouTube, such a move is necessary if we wish to further understand the complex trajectories audiences take towards, around or even away from texts in our hyper-connected and algorithmically generated digital world.Furthermore, with the increasing presence of automatic translation services on sites such as social media platforms, algorithmic mediation of the content we are shown by platform holders, and the increasing prevalence of computer-generated texts, whether in translation or not, understanding the ways in which paratextual spaces overlap cultural and linguistic borders, as well as the audience's trajectory therethrough, is of increasing importance.The main implication of collateral paratextuality for translation studies, then, is a shift away from siloed examinations of paratexts created in relation to a source or target text, to instead focus our attention on the translingual and transcultural trajectories audiences take across our hyper-connected, digital world.Indeed, to remain focused solely on the "consciously crafted" would be to remain shackled to questions of paratextual ontology and auctorial authority that fail to fully acknowledge the complex ways in which audiences move through paratextual spaces, as well as the ever-changing nature of the spaces themselves in contemporary digital contexts.

Notes
1. Indeed, earlier in her book Batchelor notes that "scholars are divided over whether underlying features such as code should be viewed as paratexts" (2018, 50, see also 57, 65). 2. The version as analysed here has been archived in the WayBack machine and is available at the URL: https://web.archive.org/web/20220714113653/https://www.youtube.com/watch?v=ESEUoa-mz2c.3. Competitor sites such as Dailymotion, for instance, do not permit users to comment directly on videos in the same way as YouTube.

Figure 1 .
Figure 1.A screenshot of the personal recommendations I received when viewing the teaser trailer for Dark on YouTube.

Figure 2 .
Figure 2. A screenshot of the recommendations when viewing the teaser trailer for Dark on YouTube in "private browsing" mode.