Following up on my previous article about deleted scenes, this piece discusses the existence of ‘inferred versions’ of films. Among other things, it defines what ‘inferred versions’ are, details their connection to the process of textual revision, how one can actually deduce that these versions existed at some point in the text’s internal history.
Deleted Scenes and Inferred Versions
To reiterate some points made in the previous article, deleted scenes function as evidence of the fact that a film text has changed over time through the process of editing. Watching them allows the audience to infer that other versions of a picture existed in the past or potentially could exist in the future.
In pointing us to the act of film revision, they evince the existence of what textual scholar John Bryant has termed “inferred versions,” which refers to alternate versions of a media text, be it a novel, film, television episode, videogame, etc. that are not available in material form to audiences, but whose existence can be confirmed by available trace evidence.
As Bryant puts it, inferred versions are
“…historically real even though they can only be speculated upon and constructed out of partial evidence found in the document… These historically constructed versions have a reality, even if they are only partially represented or inferred in the document.”1
Bryant’s writing on inferred versions is admittedly somewhat confusing. It is not always clear, as to whether or not versions of a text that had once physically existed in the past but no longer exist in the present count as inferred versions, or if the term refers solely to versions that had never physically existed but can be considered real due to evidence of revision and the conception of such a version having taken place.
To clarify things, I would modify Bryant’s concept to distinguish between “inferred real versions” and “inferred imaginary versions.” The former will refer to those versions of a text that existed in some form in the past, but are no longer fully available in the present. Meanwhile, the latter will refer to those versions of a text that does not exist in full but whose existence can be imagined thanks to the available evidence.
For instance, the incomplete set of deleted scenes a cinephile locates on a DVD may have been part of an earlier complete workprint version of the picture that existed in the course of its making. But a version containing only this specific set of deleted scenes is not likely to have been made by the filmmakers at any point in the production.
Nonetheless, the second version can be inferred by audiences. Indeed, the presence of such scenes on the DVD arguably compels such inference.
From this perspective, deleted scenes simultaneously clue the audience to at least two inferred versions: one that has actually existed in the past and one that exists in the imagination. [This also is not limited to movies; can also be applied to TV shows and videogames.] These two versions can have many overlaps but also considerable differences. And the distinction between them is not always obvious.
Deleted Scenes, however, are not the only pieces of trace evidence that allow one to make inferences about what earlier versions of a film text looked like. They function in concert with many other sources or, more specifically, other paratexts, especially advertising and behind-the-scenes materials. These include:
The Text Itself
A film can sometimes be akin to a palimpsest, by which I refer to
“a manuscript or piece of writing material on which the original writing has been effaced to make room for later writing but of which traces remain.”
That is to say, there are sometimes films or versions of films released that contain obvious and visible traces of their own alteration and/or whose compositional elements that are typically meant to be imperceptible, such as the editing, narrative, sound, or music, attract attention to themselves, destroying a viewer’s immersion in the process. Of course, the extent to which one may notice such things depends on the individual viewer. And yet, I’d argue that even when one doesn’t exactly notice such elements, they can still impact one’s viewing experience, for they fail to provide a sense of unity.
One can consider Casino Royale (1967), Highlander 2 (1991), Fantastic Four (2015) and the theatrical Justice League (2017) as examples of what I would describe as “palimpsest films.” Each theatrical and ostensibly ‘complete’ release has obvious seams in the editing and storytelling, creating the impression that disparate pieces from different versions of the picture were cobbled together haphazardly.
Sometimes, you can tell that something was cut, due to abrupt transitions in the music or sound. Sometimes, you can glean that a scene was reshot due to differences in the same scene between the appearances of the characters. Sometimes, you might become confused by what are pretty large gaps or holes in the story.*
*For instance, in Justice League Henry Cavill’s Superman sometimes has a very different face from one shot to the next in what is supposed to be the same scene.
Screenplays
Many screen texts begin with the script, meaning most, if not all, the scenes we can find in a movie or TV episode exist in some earlier written form prior to us being able to view it in a ‘finished’ form. Because of that, reading a script and comparing it with the ‘final’ product can provide a great deal of insight into how a screen narrative has changed over time, as well as make us aware of what versions of the text may exist or had, in fact, existed at one point but never made it to the screen. Arguably, the best source for comparison is the final shooting script.*
* One thing to keep in mind is that film studios far from always release all (if any) deleted scenes a given movie has on video. Therefore, in lieu of viewable deleted scenes, the final shooting script can be the only means for those interested to learn what a film was intended to be, as opposed to how it ultimately turned out.
As previously established, the final draft of the script can be considered the first cut of the movie. And this is true, in that it essentially represents the closest thing to a transcript of the first assembly cut of a film, one that lets readers know what scenes, characters, concepts, ideas, etc. existed or were intended for a film’s narrative prior to on-the-fly changes in filming, post-production, reshoots, and so on.
Trailers
Trailers (those commercials that advertise new movies) are some of most accessible and popular sources for version inference. As the case with Spiderman Homecoming demonstrated, it is common for trailers to contain deleted and alternate scenes that are otherwise not accessible to audiences. On top of this, trailers can demonstrate differences in audio, lighting, color, and digital VFX from the released text.
These can range from the minor cosmetic changes to such radical alterations that the trailer could be argued to misrepresent the final product. The fact is that marketing for movies generally begins long before they’re finished.
In this regard, the paradox is that trailers are supposed to provide audiences with an impression of the end product they will get later, yet they are based on a work-in-progress that is subject to change. In some cases, the reception of the trailer itself can lead to changes within the film text proper.
When the trailer for the videogame adaptation Sonic the Hedgehog (2020) premiered at the end of April 2019, there was a series of negative reactions towards the design of the titular character, who was a wholly digital creation. This led to the filmmakers postponing the film’s release date in order to revise the Sonic design.
This means that an unfinished alternate version of the picture featuring the initially intended Sonic design exists, and that an imaginary completed alternate version with that initial design can be inferred.
(Re)Casting Announcements
Casting announcements are a regular part of the entertainment news cycle. As such, they provide considerable insight into the making and evolution of a picture.
Every once in a while, an actor is cast in a film, only to have been completely cut prior to release. The deleted footage is not made available, yet news of the actor’s appearance in the picture remains, allowing one to infer an alternate version of the picture with the actor in it. Such was the case with Jessica Henwick, due to star in Godzilla vs. Kong, whose character was removed from the picture entirely some time in the course of production and/or post.
A similar yet different instance of this occurs when a character in a movie is recast after scenes have been filmed, pointing to the existence of a (possibly incomplete) version of the picture with a different iteration of said character.
A recent example is All the Money in the World (2017, dir. Ridley Scott).
The film originally starred Kevin Spacey as aged billionaire John Paul Getty and a theatrical trailer featuring Spacey came out in September 2017. A couple of months later, Spacey was accused of sexual assault, which led to Scott making the decision to recast the role with Christopher Plummer.
In turn, this required filming about 400 new shots featuring the character, a process detailed in this article by The Hollywood Reporter. Footage of Spacey-as-Getty appears in the initial trailer, which means a nearly complete version of the picture with Spacey exists, though it does not at the moment appear that it will ever be released.
Behind-the-Scenes Materials
BTS materials, especially those appearing on home video, such as making-of documentaries and articles, production notes and audio commentaries, are a central source of information in regard to how a film transforms in the course of its making.
In my opinion, the best bts sources for version inference are interviews - be they audio, video, or print - with cast and crew members that appear in the popular press. Interviews with film directors in particular tend to reveal the existence of alternate earlier versions that existed at one point or another in the filmmaking process.
For example, interviews conducted by Steve Weintraub of Collider.com director Joseph McGinty, better known as McG, on the subject of the 2009 film Terminator: Salvation indicate that multiple unreleased versions of the picture exist.
In a January 2009 interview that had been conducted while the film was still in post-production, McG tells Weintraub (in response to whether or not a running time for the film has been reached) that the picture is still being edited yet an expected duration is 130-135 minutes.*
* The picture’s final theatrical release, however, would have a running time of 115 minutes.
In a later May 2009 interview posted close to the picture’s release, McG apparently claims that the Home Video Release of Salvation will include 30-40 minutes of deleted scenes. McG insisted that 30-40 minutes would be present on DVD and Blu-Ray.
Despite this, the 2009 video releases did not feature any deleted scenes but did include a so-called “director’s cut” that was about 3 minutes longer than the theatrical. Then, in a February 2012 interview made for the release of This Means War, McG apparently mentions at the end that he had those 30-40 minutes of deleted scenes “in the can” but is “not sure when they might see the light of day.”
The information McG has shared suggests that his initial director’s cut that was 130-135 minutes long and that the picture was cut down by 20-25 minutes for theaters. Furthermore, the director’s repeated claims about 30-40 minutes of deleted scenes point to the potential existence of an even longer version of 160-175 minutes, as well as evince that Warner Bros. was initially going to make the deleted footage public, only to change its mind and pull it from the video releases.
His claims were supported by the availability of other behind-the-scenes materials, including concept art and VFX reels, that provided glimpses of the unreleased scenes.
Inferring Versions and (Re)Constructing Histories
All in all, paratexts such as deleted scenes allow one to infer the existence of unreleased alternate versions, both real and imaginary. This is because they usually contain disparate pieces of information that tie into a larger historical narrative of the film text’s transformation. In other words, paratexts can provide great insight in the internal history of a specific film. But more than that, they function as proof that motion pictures qualify as examples of what Bryant terms the “fluid text.“
The author uses the term to emphasize the idea that variation is inherent to a given work or text. As he puts it:
“Simply put, a fluid text is any literary work that exists in more than one version. It is “fluid” because the versions flow from one to another. Truth be told, all works – because of the nature of texts and creativity – are fluid texts. Not only is this fluidity the inherent condition of any written document; it is inherent in the phenomenon of writing itself.”2
In other words, a media text fundamentally cannot exist in a single, stable version. Instead, it is always subject to change and revision.
I fully agree with this viewpoint, and hope to further illustrate how fluid film texts are in this and other articles. In the process, I hope to change the predominant cultural perception that movies are unstable, unchanging works of art.
If you are interested in the subject further, I’d suggest reading the following two articles I’ve found, which exemplify how one can infer versions from existing evidence.
If you like this article, then please, by all means, share it!
Want to learn more about alternate cuts, versions, editions, etc., or just read some interesting film and TV criticism? Then sign up for my newsletter! You will get free Substack-original articles as well as free links to my paywalled essays on Medium.
John Bryant, The Fluid Text: A Theory of Revision and Editing for Book and Screen (University of Michigan Press, 2002), 79.
Bryant, The Fluid Text, 1.