Reading a comic book is easy–even when there are no words to be read. You just look at a picture, and then at the next picture, and so on. But why do any of them make sense side-by-side? What is your brain doing as it leaps from image to image?
In Understanding Comics, Scott McCloud defines the Gestalt psychology principle of “closure” as the “phenomenon of observing the parts but perceiving the whole” (though it more specifically indicates a viewer filling in visual gaps between disconnected parts) and applies it to comics gutters: “Nothing is seen between panels, but experience tells you something must be there!” He goes on to explain: “Comics panels fracture both time and space, offering a jagged, staccato rhythm of unconnected moments. But closure allows us to connect these moments and mentally construct a continuous, unified reality.”
McCloud focuses his analysis on gutters and therefore types of transitions possible between panels (though closure is independent of panels and gutters, since insets and interpenetrating images work the same ways). He comes up with six types:
They work reasonably well, but his focus on panel transition has always struck me as slightly off. When I use it in class, students often don’t come to a clear consensus when analyzing any particular panel sequence. Moment-to-moment and action-and-action, for instance, are often ambiguous, sometimes combining identical leaps in time. And since actions do occur in McCloud’s moment-to-moment examples (a women blinks!), it’s not exactly clear what constitutes an “action.” Aspect-to-aspect can also be indistinguishable from subject-to-subject, both of which may or may not involve a movement in time, and so may or may not also be moment-to-moment or even action-to-action. And scene-to-scene might be a location leap and so also a kind of aspect-to-aspect at the big picture level, or a scene-to-scene can be in the same location but at a different time–so then how much time has to pass for an old scene to become a new scene?
These are annoying questions, but they really do come up when you try to breakdown a panel sequence with a roomful of students. So instead of categorizing transitions, my colleague Nathaniel Goldberg and I categorized types of closure while drafting our essay “Caped Communicators: Conversational Depiction and Superhero Comics.” Instead one all-purpose “perceiving the whole” process, we see four very different kinds of closure, each of which can occur by itself or in combinations.
Spatial: Subject matter drawn in separate images is depicted as existing in physical relationship to each other, typically as a result of panel framing. (What McCloud identifies as aspect-to-aspect, subject-to-subject, and some scene-to-scene transitions require spatial closure.)
Temporal: Undrawn events are depicted to take place outside of events drawn in separate images, typically as a result of panel transitions and so occurring as if in gutters. (What McCloud identifies as moment-to-moment, action-to-action, and some subject-to-subject and scene-to-scene transitions require temporal closure.)
Causal: Drawn action is understood to have been caused by an element absent from a current image but drawn in a preceding image. (None of McCloud’s transitions, not even action-to-action, accounts for this type of closure.)
Associative: A metaphorical relationship is depicted between two images in which one image is understood to represent some idea about the other image. (Though McCloud does not identify this type of closure, Jessica Abel and Matt Madden in Drawing Words Writing Pictures add “symbolic” to McCloud’s list of transition types. Such symbolic transitions require associative closure.)
It always helps to look at specific examples, so consider this three-panel sequence at the top of page 28 in Watchmen #8:
In the first image, artist Dave Gibbons draws the shadow of a statuette cast over the face of a frightened man lying on the floor. The second image shows the statuette in the fist of an attacker. Taken together, spatial closure is required to understand that the two images occur within a few feet of each other, each image drawn from one of the two men’s points of view. The second image also requires temporal closure because the statuette is behind the attacker’s head at an angle that would not cast the shadow seen on the victim’s face in the first image. Gibbons therefore also depicts a movement forward in time during which the attacker has cocked his arm back to strike.
The third image depicts a jack-o-lantern crashing to the floor with some falling books. It uses all four forms of closure. The pumpkin exists in the same space as the two now undrawn men (spatial closure). The pumpkin is crushed at a moment immediately following the second image (temporal closure). The falling books have been knocked down by the now undrawn attacker of the previous image (causal). And, because it resembles a human head and breaks open at the moment a reader anticipates the statuette striking the man’s head, Gibbons implies that the man’s head has been similarly damaged (associative).
A close reading of the sequence also reveals some confusion. Regarding causal closure, it is unclear how the attacker overturned the books at this moment since the act of swinging the statuette at the victim on the ground does not clearly involve his intentionally or unintentionally knocking over the bookcase in the same gesture. Instead, Gibbons may have attempted to suggest that the attacker struck his victim and then afterwards overturned the bookcase—an ambiguous two-step action otherwise absent.
I’m guessing Gibbons was fulfilling a directive in Alan Moore’s famously meticulously detailed script, producing this unintended gap in its execution. To address panel transitions that cause only confusion, McCloud includes “non-sequitur” as a type of transition that “offers no logical relationship between panels whatsoever!” So then a non-sequitur produces no closure at all–and so isn’t really type of closure, but is the absence of closure. Which is why we don’t include it as our fifth category.
So our closure types are deeply indebted to McCloud, but I think they also improve on his. I’ll be testing these out in my classroom soon, so hopefully my students will agree. More on that later.
“Regarding causal closure, it is unclear how the attacker overturned the books”
The victim falls backwards into the bookshelf, right? You see the pumpkin on the shelf in panel one. So there’s a sort of closure going on between panels one and three in a way.
Yeah,I read it like Ben above. Mason is not lying on the floor, he is on his knees or getting back up from the floor and falls back when hit knocking the pumpkin and books down.
Yeah, I see it now. He’s definitely not on his back. That does clarify the final panel a lot!
Interesting to read your type of transitions, especially that the focus is on what is inferred, rather than what is depicted.
I actually proposed something similar to your Associative type (and A&M’s symbolic type) back in my 2003 book, where I both expanded the number of transitions to 8, but also provided broader categorization to them (ex. temporally progressive vs. temporally ambiguous).
This approach eventually gave way to my “grammatical” approach, because I realized that transitions just couldn’t account for lots of aspects of comprehension (like panels connecting across distances, and groupings of panels into segments, you can only make a “transition” once you’ve gotten to the next panel meaning it isn’t “in the gutter”, etc.). I discussed the problem with transitions in my books, but more thoroughly in this paper: http://visuallanguagelab.com/P/NC_Time&Transitions.pdf
Interestingly, many cognitive approaches to discourse hold a similar idea that people track different elements in a narrative (characters, spatial location, time, causation) and that you update your “mental model” of the scene as these things change across units (be it sentences or panels). This is similar to your approach. (See papers by Magliano or Zwaan about this, often called the “event-indexing model”). I’m actually on board with this idea, and have shown in experiments that this processing runs parallel to my narrative grammar.
When my students code panels for our ongoing corpus study (http://www.visuallanguagelab.com/vlrc/index.html), we code for changes between panels of characters, spatial location, time, or causation. However, these are non-exclusive (more than one can happen at a time) and non-discrete (we code for “full changes” as well as “partial changes”). We get pretty good agreement across coders usually. So, overlap with yours too.
10,296 possible transitions! That’s a great analysis, Neil, especially your point that to understand time requires understanding the images as depictions (which then gets into a rather massive philosophical area as far as defining the relationship between an image and the subject it somehow represents).
I think panel-to-panel transitional closure is still useful, but inadequate to discuss the larger narrative elements which your visual sentencing addresses so well. I also think there needs to be an additional category that’s between panel-to-panel and visual sentencing (at least when visual sentences are multi-paged).
The closure-like relationships between panels seen simultaneously on one page. Groensteen addresses this somewhat with rhetorical layout, as does Molotiu with sequential dynamism and iconostasis. But it would be interesting to examine how an awareness of the page layout (understood non-sequentially but as a whole and narratively as a kind of prolepsis) affects the sequential comprehension of each panel. For instance, in your example of the 3-panel strip ending in the dog chasing its tail, a reader sees the final image (at least briefly and peripherally) before understanding its narrative meaning when she literally reads the words in sequence. That visual preview creates a different experience than someone seeing one panel only at a time–as with many webcomix.
Anyway, very glad to hear my closure categories aren’t completely out of wack with your far wider research!
Interesting to hear your thoughts. I agree that layout, meaning, and narrative interact, but I don’t think there’s an additional binding structure there. At least, I don’t see the need for one just yet that isn’t covered by the linkage of those existing structures.
Actually, we have some data about eye-tracking in comics that are presented all at once (a 3×2 panel grid) or one panel at a time (the paper is in revision). We didn’t find that eyes jump forward too much to get a preview, though they do regress back to adjacent panels sometimes (much the same as reading text). They also don’t do a “whole page scan” as has been claimed, and mostly just stick to the linear sequence. This linear reading is true of what’s found in full comic book pages (not strips) too.
The biggest difference between layout types is that people spend more time looking at each panel when it’s one panel at a time vs. all at once, likely because they know they won’t have the option of looking back once that single panel is gone. Generally they look at the same information within panels in both though.
Happily this is where your research-based claims trump other comics theory. I was accepting based on plausibility the common claim that the page is first taken in as a whole. I’m fascinated that readers don’t appear to actually do this–at least not measurably. I wonder if peripheral awareness is still a factor though? This might partially explain why cliff-hanger panels only work at the bottom of right-hand pages. Though of course the forced pause involved in a page turn is greater than the pauses between panels, so peripheral awareness isn’t necessarily a factor.
I wonder if anticipating a panel being gone entirely explains more time looking at it though. When I read panels in a multi-panel page layout, I’m at some level aware of the overall page, and that each unit is only a portion of the larger unit, and I attend to each unit in some proportion to the whole. But if the lone image is the whole, I attend to it differently. I’m always trying to get to the panel in the bottom right corner. Pausing before that feels unnatural (like pausing in the middle of a poetry stanza, even if there’s a sentence that ends in the middle of the stanza). In comic books, rather than comic strips, the page is the primary unit of composition and, I think, consumption. Though, frustratingly, eye-tracking might not get at that theoretical page awareness.
Looking forward to reading this next paper of yours, Neil.
Thanks for these posts, Chris. They’re super-valuable for those of us without institutional access to comics academe.
One thing that occurs to me about these discussions of transition/closure: comics is not the only medium where the audience has to do some inferential work to determine the relation between two (or more) juxatposed images. It happens in film too — how are these two consecutive shots related? Surely there’s useful discussion in film studies about the processes involved?
Chris, I can’t say much about pausing within pages (the data I’ve seen doesn’t show much pausing in pages at all), but I can say that none of the eye-tracking studies of comics that I’ve seen—whether I’ve been a part of them or not—have ever shown the “people scan a page first” effect. I know Groensteen mentions some never-published highly sketchy sounding eye tracking study from France in the 1980s that (I think?) is the source of this meme, but I know of no data that supports it.
That said, I do have data that shows people make anticipations about subsequent panels. But, these seem to be based on the content in one panel allowing for predictions made in other panels. This is mentioned in my “The grammar of visual narratives” paper, which is summarized in this video: https://youtu.be/dU7c9euTso4
Essentially, this paper showed clear evidence that we *don’t* rely on backward-looking panel transitions, and we do make forward-looking predictions about subsequent panels based on the content of panels we’ve seen before.
Jones, there is definitely a useful literature from film theory, and many current people on both the film and discourse side of things have been making links with each other. The work by Magliano and Zacks applies “transitions” to both film shots and sentences (and comics). Zacks has a very readable book called “Your brain on film” about this work. Some semiotics work on film (with no connection to cognition) has recently been applied to comics too. I believe my theory of narrative grammar works for film too, though it gets modified due to the moving of the images (currently working on a book chapter about this).
Yeah, and now that I think about it, Neil, my hypothetical race-to-the-bottom-of-the-page tendency doesn’t occur when reading prose-only texts and so there’s little justification for assuming it occurs with a comics page either.
I think the scan-the-whole-page idea may come organically from how comics are drawn one page at a time. Since the page is the unit of composition, it makes intuitive sense that the page would also affect reading somehow. But apparently not.
Do Magliano and Zacks apply transitions to prose sentences, and so the “gutter” between a period and the capitalized letter of the next sentence? If so, I really gotta read that. I’m increasingly interested in the application of comics theory to prose.
Well, I wouldn’t characterize Magliano and Zacks’ approaches as “transitions” necessarily, and definitely not with a “gutter” between sentences. (I find the emphasis on the “gutter” as somehow meaningful in comics as fairly ridiculous actually). Rather, they focus on how meaningful changes from sentence to sentence alter one’s conceptualization of the overall scene. The approach was made for discourse, but has subsequently been applied to film (and kind of comics). The main reference for this theory is here:
Zwaan, Rolf A., and Gabriel A. Radvansky. 1998. Situation models in language comprehension and memory
Yes, “gutters” are imaginary. The notion that the undrawn white space within a panel frame is somehow different from the undrawn white space outside a panel frame is ridiculous. For that matter, panel frames aren’t actually “frames”–they’re representations of frames. The only frames in comics are page edges. I was recently arguing these points in an essay I have under submission at the Int. Journal of Comic Art. Emma Rios’ art in Pretty Deadly provides a great example of how gutters and frames are irrelevant categories for comics theory.
Anyway, thanks for the Zwaan reference. I’m looking it up now!
The notion that the undrawn white space within a panel frame is somehow different from the undrawn white space outside a panel frame is ridiculous. –
…but of course it is different- they are the same white space signifying different things. It’s like saying the dot in the i in ‘different’ means the same thing as the dot at the end of this sentence. Don’t forget comics is not writing, comics is drawing.
You say it much better, Ibrahim:
“they are the same white space signifying different things”
That’s exactly what I was trying to say. Sometimes the gutter is talked about as if it were a formal element or even a physical structure and not areas of canvas that the artist leaves white to signify something. The gutter isn’t neutral or empty any more than an undrawn area signifying clouds is neutral or empty.
Well, put like that i can indeed find no fault with the statement. However, that makes it *exactly* a neutral space, to be activated by the artist’s intentions. The white spaces between Miller’s panels in the Dark Knight or Brown’s Paying For It are closer to that original neutrality than say, Miller’s use of them in R?nin, i’d argue.
Your posts constitute important work, by the way, in forcing us to consider these aspects.
Thanks so much, Ibrahim. And, yes, ALL white space on a page canvas begins as neutral space and then is activated, whether drawn on or left undrawn.
I’ve also noticed that undrawn gutters aren’t as popular as they used to be. They’re still around, but Convergence, for example, uses mostly insets and overlaps instead.
“Sometimes the gutter is talked about as if it were a formal element or even a physical structure and not areas of canvas that the artist leaves white to signify something.”
is there a popular example of this in comics scholarship? or are you saying this is how people that come from outside the comics world would talk about gutters?
Tim, I’m talking about comics scholars. Here’s an example from two philosophers, typically the most exacting of definers:
Hayman and Pratt define a comic as “as a sequence of discrete, juxtaposed pictures that comprise a narrative, either in their own right of when combined with text” (423). They clarify that “the visual images are distinct (paradigmatically side-by-side), and laid out in a way such that they could conceivably be seen all at once. Between each pictorial image is a perceptible space; we’ll call this the gutter” (423).
And the best known example is from Scott McCloud, the grandfather of comics scholarship, quoted above.
Chris, of course I know McCloud, though I never actually read Understanding Comics as I just dont like the way he draws. Im not sure I get the distinction you’re making between what people think gutters are and what gutters actually do … but maybe i just need to read your article and comments again more closely. Anyway thanks for your response!
Re philosophers defining things, and Hayman and Pratt in particular, have a look at Aaron Meskin’s “Defining Comics?”. It in the Journal of Aesthetics and Art Criticism in 2007.
I don’t recall it discussing gutters in particular, but it’s pretty savage on the McCloud and Hayman-Pratt (and all of the other) attempts at defining comics.
Yes, I’ve got Meskin right here on my desk too, Scott. He gets some things wrong too though. Here’s a snippet from another article I have under review:
Meskin also criticizes the anachronistic or ahistoric use of “comics” to include artworks, such as the 9th century Bayeaux Tapestry, that predate the coinage of the term. By such logic, Jules Verne and H.G. Wells are not science fiction authors because “science fiction” was coined in 1929. According to the Oxford English Dictionary, “chiaroscuro” dates to 1686, after the deaths of its exemplary artists, Rembrandt and Caravaggio, and the technique is commonly traced to Ancient Greece. Ahistoric terminology is a norm of categorization and analysis.
” Ahistoric terminology is a norm of categorization and analysis.”
I wouldn’t say ahistoric…or I guess I’d say that there’s nothing ahistoric about talking about H.G. Wells as if you can read him and/or put him on a bookshelf now. Artwork looks different from now, in part because of how genres develop, and so it makes sense that you’d look at War of the Worlds differently from here than from there. Part of that is seeing how it fits into current genres, which were influenced by it, and now influence our reading of it.
Chris, I read Meskin a bit differently on the anachronism point.
He agrees that it might sometimes be useful to close a historicist concept for certain analytical purposes. One example would certainly be the analysis you are giving of the way statically arranged images narrate.
But Meskin’s main argument is that Hayman and Pratt (and everyone else) fail to get remotely close to our pre-theoretic concept of comics. Everyone agrees that Caravaggio used chiaroscuro and just about everyone agrees that Jules Verne wrote SF. But the Bayeaux tapestry is not like that. No one outside of the defining comics discussions calls _that_ a comic.
I would say that the better analogy to Verne is someone like Topffer. The Bayeaux tapestry case reminds me of people who claim One Thousand and One Nights as SF. There are elements of each older work that are similar to the relevant modern category, and thinking about that can illuminate both sides. But that’s no reason for a wildly revisionist definition.
It’s a tricky point. I do think some pre-comics art can reasonably be called “comics” if analyzing them formally as comics yields useful results. The other term is “image-text,” which gets less resistance I think.
Here’s more of my essay draft on that topic:
If comics are the art form of juxtaposed images, then a great many artworks that are not typically or historically considered comics must apply. If comics are instead a subcategory within the wider field of juxtaposed image art, then the Bayeaux Tapestry remains apart, and comics scholarship is limited to a specific subset of published works. Brian Attebery’s system of division for fantasy prose fiction into formula, genre, and mode may be further instructive. Comics as a mode might include all examples of image juxtaposition (including advertisements and road signs); comics as a genre might include a smaller but still diverse set of art (including the Bayeaux Tapestry and Matisse’s Jazz, as well as the engraved poetry of William Blake and the word and photography-combining art of Barbara Kruger); and comics as formula only those works published explicitly as comic strips, comic books, or graphic novels and memoirs. All formula works would be contained in the comics genre, and all genre works would be contained in the comics mode.
The definition of comics derived from all meanings available from word and picture combinations is the most expansive and so therefore describes the comics mode or word-picture method—and so then it is not a body of works per se. Adjusting Atterbery’s terms, comics as genre may be better termed the comics medium, the inclusive body of artistic works that apply the comics method. And, at the risk of confusion, comics as formula may be better termed the comics genre, those works that identify themselves within the publishing tradition of graphic narratives.
That’s a lot to chew on. I look forward to doing so tonight.
OR… We could acknowledge what I’ve been saying since my 2003 book and 2005 paper “Undefining comics” that comics are not defined by any specific combination of images (in or out of sequence) and text. (Repeated in my recent book: http://www.thevisuallanguageofcomics.com)
Having a notion of a visual language that “comics are written in” divorces “comics” from being dependent on “comics medium.” This is just the same as novels or magazines. We wouldn’t say that novels or magazines ARE a written language, but that they are written in a written language. Comics likewise are written in a visual and/or written language, and are defined by a lot more than just text and images.
Comics-as-method is a good paradigm. It helps to get a straight picture of the debts owed. For instance in a gallery setting, when works are positioned next to one another with only a general idea of their juxtaposition, this is just standard practice; but when each of these images is positioned with a specific juxtaposition to the others in mind, then it can be said that comics-as-method is being applied; this debt goes unacknowledged too often, Lichtenstein-style, and this surreptitiousness (or ignorance) prevents the method from being seriously studied by practitioners of all visual arts. There’s no need to look that far back to Bayeux; when Anselm Kiefer is using writing in his paintings there is Wilhelm Busch to consider; Damien Hirst’s shark exists in a perpetual split-panel breakdown, &c.
Chris, one thing that I think might help (or at least help me) is to separate two different senses of “definition”. One, which I take Meskin to be talking about, is being precise about what we already mean by “comics” outside of the seminar room. And that one is easy — it’s your comics-as-formula. This makes your passages opening — “If comics are the art form of juxtaposed images” — grate, because comics are not that.
But we might also want to stipulatively define [term] to play an analytical role in a specialized setting — e.g., something that covers anything that can be (fruitfully) formally analyzed as a sequence of images. The question of whether to replace [term] with “comics” or “image-text” is a tricky one. I can convey my own position here by abruptly changing topic.
I am an economist by training. My technical vocabulary contains a bunch of special words that hide as ordinary English words. Examples are “cost”, “profit”, “competition”, “efficiency”, and “equilibrium”. None of those mean, inside of economics, what they mean in the ordinary English of people in other disciplines, or in journalism. And that is the root, I think, of a lot of miscommunication in the translation of economics into popular discourse. So I often wish we had come up with something ugly and strange to play the role to “cost” that “image-text” plays to “comics”.
All of that said, I think there is actually a lot of agreement here, obscured at first by my reflexive hatred of _Understanding Comics_. But let’s not talk that through right now…
Neil, thanks for chiming in. That paper makes me want to order your book. In particular, I think this is exactly right, and is a way of cashing out some of what I said earlier: “It must be remembered that “comics” is an artifact bound to its socio-cultural context, and cannot be extended as a pan-temporal and cross-cultural universal – an assignment that is available to the structure of visual language.”
When your book percolates up to the top of the unread books pile, I get an opinion of “visual language”. I take it from your reference to being literal about “language” you are not really working in the semiotic tradition that used to be so important in film studies?
Scott, your point about economics terms is persuasive.
In which case, adjusting Attebury’s 3-step progression, we might get:
1. image-text method
2. image-text art
3. comics
That probably works better. Though I admit a part of my perverse soul likes applying the word “comics” to all three. And, as Ibrahim points out, “comics” is useful there in other ways–especially when looking at wordless image juxtapositions.
Chris, I am definitely working outside of the semiotic tradition. I most certainly mean “language” in a literal way, consistent with contemporary linguistics and cognitive science, particularly the work of linguist Ray Jackendoff (as in my recent paper that integrates “visual language” into his cognitive architecture for language: http://visuallanguagelab.com/P/NC_multimodality.pdf).
I usually break it down somewhat like this:
A person speaks in a spoken language. Spoken languages differ across the world, such as English and Japanese.
A person draws in a visual language (assuming they’ve acquired one). Visual languages differ across the world, such as American VL and Japanese VL.
“Comics” are usually written in both a written language (spoken mapped to visual form) along with a visual language. The specific visual language that is used is often highly associated with particular types of comics. Manga are most often drawn in Japanese Visual Language (whether or not the author is Japanese) while superhero comics are drawn in “Kirbyan” American VL.
BUT… illustrated kids books and instruction manuals also use text in combination with visual languages. They are NOT “comics.” But, they may use different visual languages than the ones in comics.
The specific types of visual languages factor into whether they are called “comics” as do a wide range of sociocultural factors, including sociocultural context, intended demographics, format, etc.