Social Network Analysis of co-occurring comic characters

Another thing I learned at my librarian job is that Social Network Analysis (SNA) methods seem to become increasingly popular in the Humanities. The basic idea of SNA is that you define a type of entity as nodes (actors), and some criterion for establishing edges (connections) between them. Once you have constructed such a network, you can analyse it by applying various mathematical operations. The difficult part is defining your nodes and particularly your edges in a way that is both feasible and meaningful.

Some Literature scholars have tackled this problem by using SNA for drama. Written plays are highly structured: speakers are indicated in fairly standardised ways, so that they can be used as nodes in a network. Edges between them can be formed by looking at which characters are on stage at the same time (i.e during the same scene), possibly indicating a dialogue or other interaction. Another benefit of using drama for SNA is that many older texts are available digitally. Crowdsourcing may be used to clean up this data, thus making it machine-readable for SNA purposes. The resulting graphs may provide insight into certain historic developments, e.g. the number of characters per play increasing over time (PDF, German).

In comics, such automatic processing is still a distant dream, but on a smaller scale, networks may be constructed manually. Identifying nodes is more problematic in comics, though, because unlike in drama, characters aren’t explicitly named each time they appear. They usually have to be identified by their looks, which isn’t always easy. Another problem is how to define the edges. A research group from Paderborn recently proposed (PDF, German) to establish an edge between two characters whenever they appear on a page together. In my opinion, a more suitable category than the page would be the panel, as there are sometimes narrative shifts between panels on the same page, so that the co-occurrence of characters on a page doesn’t necessarily imply interaction. Furthermore, some comics don’t have pages, but they all have panels.

To test the feasibility of this approach, I built a little character network based on co-occurrence within panels, once again using Akira. Here is a Gephi rendering of such a network from the first 16 pages of volume 3 (blue numbers indicate the number of panels on which both of the connected characters appear):

Character co-occurence network of Akira vol. 3, pp. 5-20I assigned the group of soldiers to one single node rather than one node per visible soldier, similar to a speaker designation for groups of people in a play. As we will see in the second example, these ‘crowd’ nodes may cause some headache. Anyway, the most striking thing about this network is that it consists of three unconnected clusters. In other words, the action takes place at three different places on these 16 pages: the military base, Miyako’s temple, and the streets of Neo Tokyo. (Actually there are two more locales – the site of the SOL laser beam impact and SOL in space – but no character interaction takes place there.) Keep that in mind as we look at the first 17* pages of the 4th volume:

character co-occurrence network of Akira, vol. 4, pp. 4-20

2 panels from p. 13 of Akira vol. 4 by Katsuhiro Ōtomo

Spot the Lieutenant.

At first glance, this graph is very different from the first: instead of three clusters, there is one small and one large cluster. However, this impression is misleading. Because I lumped all of Tetsuo’s henchmen together as “Great Tokyo Empire mob”, they act as a bridge between the actually unconnected scenes at the rescue helicopter on the one hand, and Lieutenant Yamada and his diving unit entering the city on the other. (Another problem here is that Yamada can’t be recognised until he takes off his diving suit – for simplicity’s sake I just assumed he is always among the group of divers depicted.)

Thus we can tentatively recognise a pattern in Ōtomo’s storytelling: rather than building his story around one central protagonist, he frequently jumps between parallel lines of action, with shifts taking place approximately every 2-8 pages.

*Why the different number of pages (16 and 17, respectively)? The reason is that I analysed both volumes until p. 20, but vol. 3 starts on p. 5, whereas vol. 4 starts on p. 4.

Some notes from the 33rd Kunsthistorikertag, Mainz, March 24-28, 2015

photograph by Peter Pulkowski from

Kunsthistorikertag, the conference of the German society of art historians (VDK), takes place every two years and is one of those huge events with hundreds of participants and up to four parallel tracks. I won’t even try to sum up all the talks I’ve heard. Instead I’ll just pick one ‘highlight’ from each of the five conference days with some connection to sequential art and/or Japan.

  • Tuesday: One of the three parallel talks that kicked off the conference was Miguel Taín Guzmán on “The views of the cities of Spain drawn by the Florentine artist Pier Maria Baldi [ca. 1630-1686]: the codex of the journey of prince Cosimo III of Medici in the Laurenziana Library” (one of the few presentations in English by the way). The codex consists of a written account of the prince’s journey, interspersed with 86 ink drawings of cities and other stations of the journey. Some of these drawings not only depict the place, but also a specific situation at the time of the prince’s visit: Cosimo’s travel procession, camels at Aranjuez, a thunderstorm over Santiago de Compostela, etc. In this huge book of 60×100cm, the drawings are arranged on double pages with two drawings on top of each other. As the placement of the drawings corresponds to the chronology of the journey, one could speak of juxtaposed images in deliberate sequence… Some of the digitised drawings can be seen at, and there’s also a huge PDF of a book reprint.
  • Wednesday: Irene Schütze examined the exhibitions of Jeff Koons, Takashi Murakami and Joana Vasconcelos at the Palace of Versailles in 2008, 2010, and 2012, respectively. It was the only talk at the conference, at least of the ones I heard, that mentioned manga: apparently, Murakami “learnt about Versailles through the girls’ comic book Rose of Versailles” (Riyoko Ikeda’s ベルサイユのばら / Berusaiyu no bara, also known as Lady Oscar), a quote (published in various places, e.g. The Independent) which has been used against him by his critics.
  • Thursday: Christian Berger re-introduced the almost forgotten conceptual artist Douglas Huebler (1924-1997). Many of his works consist of photographs accompanied by writing (or vice versa?). Compared to other conceptualists, the form in which he presents his images and texts is much more important and might even be considered as the artwork itself, not merely a documentation of some photography performance as the ‘actual’ work. Location Piece #5, for instance, is a series of ten pictures of patches of snow next to a highway, taken on a road trip in specific distances, as the text explains. Talk about juxtaposed images in deliberate sequence again.
  • Friday: Helmut Leder and Raphael Rosenberg presented results of someone else’s study which tracked eye movements of people from Austria and Japan who where looking at a digital reproduction of the same painting. The paths of the eye movements and the areas on which they focused were similar within the group of Austrian participants on the one hand and within the group of Japanese participants on the other hand, but there were notable differences between the Austrian and the Japanese average. For Rosenberg, this proves the cultural influence on perceptual physiology.
  • Saturday: On the last day of the conference, several field trips were offered. I joined the one to Tadao Andō‘s sculpture museum in Bad Münster am Stein. It consists mainly of Andō’s signature concrete blocks with their characteristic hole pattern, but the building is unusual for him in that it combines an old, relocated half-timbered barn with the new concrete parts. The guide at the museum suggested the concrete blocks were modeled after tatami mats, which might be true for their measurements (90×180cm), but the pattern in which they are arranged in the museum walls differs from traditional Japanese tatami floor layouts.

Author dictionaries and lexical analysis for comics

Every once in a while I learn something at my day job that I think would be applicable to comics research too. For instance, in literary studies, dictionaries are compiled that contain all the words (or only the nouns, similar to an encyclopedia) used by a particular author, or even only those used in one single literary text. Think of it as a sort of commentary in a critical edition which explains references to real-world entities, or obscure words that aren’t used anymore, only separate from the source text and organised alphabetically.

Applying this method to comics, we would, of course, ignore all the images and lose the information they convey. On the other hand, looking at the words alone might yield interesting results. For instance, by comparing the frequency of words used in a particular comic to the frequency with which they occur in written language in general, we could test common hypotheses such as “author X uses word Y a lot”.

For comics of more than a few pages length, it would be nice to automatically create a list of all the words in digital form (at least those in speech/thought bubbles and captions – sound effects and inscriptions/labels can be difficult to automatically recognise). Unless a script for the comic you’re interested in is already available, a straightforward (though not necessarily easy) way to get such a list would be to obtain digital images (e.g. scans) of the pages of the comic, then run Optical Character Recognition (OCR) software on them.

As an example, consider these two panels from Akira, in which a scientist is introduced to some colleagues:

two panels from Katsuhiro Otomo's AkiraThe OCR software recognises the text in the five speech bubbles like this:

  1. 初めまして
  2. スタンリー・
  3. よろしく
  4. ジョノレジュ
  5. 初めまして

As far as I can see, only two mistakes (ノレ instead of ル and ですノ instead of です) were made. Instead of focusing on nouns (for which there probably are detecting algorithms for Japanese), it’s easier for now to just look at the kanji and filter out all hiragana and katakana characters. (While you can’t simply say that kanji represent nouns and kana represent other parts of speech, the idea here is that kanji tend to carry more semantic information than kana, which are often only used as flection suffixes.) That leaves us with the six kanji , 名, 前, 博, 士, and 初 again.

We can look up their frequency with which they occur in Japanese language in general, e.g. the frequency rank at WWWJDIC:

  • 前: 27
  • 初: 152
  • 名: 177
  • 士: 526
  • 博: 794

i.e. 前 is the most frequent of the five, 博 the least frequent. Compare these ranks to the frequency with which they occur in our slim sample of two panels:

  • : 33% of all kanji
  • 前, 名, 士, 博: 17% each

What we can see here, if anything, is that two kanji, 士 and 博, are significantly more often used by Katsuhiro Ōtomo than by the average Japanese author. This doesn’t come as a surprise, as the compound 博士 signifies the academic title ‘Dr.’, which is the appropriate form of address for the scientists in this scene, whereas the other kanji 前, 初 and 名 are linked to names and introductions in general, and thus more often used in standard Japanese.

However, even if the frequency of 士 and 博 remained above-average if we analysed all of Akira‘s over 2000 pages, that wouldn’t necessarily mean we had discovered a lexical characteristic of Ōtomo’s writing style. What it would tell us is that there is a subplot about scientists in Akira. Of course, topic analysis based on word frequency is nothing new. More interesting from a formal-lexical point of view would be if we discovered kanji used in different frequencies than we would expect with regard to the subject matter treated in Akira. In this situation it might be useful to look at synonyms: when Ōtomo had several options to express the same thing, why did he choose some words over others?

panel detail from Akira by Katsuhiro ŌtomoFor instance, on the same page as the example above, the relatively infrequent (rank 920) kanji 栄 is used as part of the word “honour” in the expression “I’m honoured to meet you”. Instead, Ōtomo could have used the phrase “nice to meet you” for a third time, using the kanji 初 again, but he didn’t. Suppose there was a significant number of further instances of 栄 in Akira, maybe that would be a formal-stylistic choice, rather than one merely implied by the content of the comic?

I’m aware that all this is very hypothetical, and that looking at just a few panels doesn’t show anything, but if I wanted to analyse a comic in this way, I would basically go on about it as described here, only with more scans. If you would like to learn more about this kind of analysis, I recommend Allen Riddell’s tutorial on “Feature selection: finding distinctive words”.

Index to “[theory] – in comics?” blog posts

Wait a minute, isn't it Women's History Month? Oh well...

Wait a minute, isn’t it Women’s History Month? Oh well…

According to the traffic statistics provided by WordPress, the series of “[theory] – in comics?” posts is one of the most popular (or rather, least unpopular) parts of this weblog. I even keep returning to them myself to remember what I read. However, I admit they’re not easy to find: even when using a tag or category, you have to scroll down a long page of posts displayed in chronological order of posting. Wouldn’t it be good to have an index in which you could see all of these posts at a glance, and in a more meaningful order? Or better still, several indices? Here you are: last name | keyword | year of publication | title | comics creator | comic title

These lists will be continuously updated.

Roland Barthes’s Rhetoric of the Image – in comics?

Already in 1964, Roland Barthes (1915-1980) published one of his best-known essays, ‘Rhétorique de l’image’¹. It is also a text that’s quite difficult to understand. Here’s what I make of it: according to Barthes, any image (except for those created by “illiterate societies”, e.g. ancient cultures) contains three messages, the linguistic, the literal (or denoted), and the symbolic (or connoted, or cultural). It should be noted that Barthes deviates from traditional communication theory by using the term ‘message’ without necessarily tying it to a sender: “the language of the image is not merely the totality of utterances emitted (for example at the level of the combiner of the signs or creator of the message), it is also the totality of utterances received: the language must include the ‘surprises’ of meaning.” Anyway, let’s look at these three messages in detail:

The linguistic message can be found “in, under, or around the image”. Such textual matter is always there, says Barthes (although I can image at least one kind of image where text may be completely absent). This linguistic message may have either (or both) of two functions: anchorage and relay. Anchorage means, the text clarifies or “fixes” the meaning of the image, which is always polysemous, by choosing some of the possible signifieds and “banishing” others. The relay function is less common and puts the text in a “complementary relationship” with the image – Barthes explicitly mentions comic strip dialogues as an example.²

The literal or denoted message consists of the objects depicted in the image, stripped of all symbolic meaning. It conveys nothing but a consciousness of the “being-there” of the represented things. (Barthes is particularly concerned with photography here, which additionally implies “an awareness of […] having-been-there” and a tension between “spatial immediacy and temporal anteriority”.) According to Barthes, little more than “anthropological knowledge” is required to recognise the objects in the image on this level, although I doubt it’s always as easy as that.

Finally, the symbolic message contains the signifieds (or ‘meanings’, colloquially speaking) of the image. In contrast to the literal message, more specialised kinds of knowledge may be required to decipher this message, but different “readings” of the same image are still legitimate. The totality of signifiers in an image, by the way, is what Barthes calls the eponymous ‘rhetoric’. It’s also noteworthy that Barthes seems to think that in some instances (e.g. advertisement photography, his principal example) we may even realise signifieds invested by the sender/creator of the image, i.e. recognise the creator’s intention (and there we have the good old sender-message-recipient model again).

Let’s see if we can read a panel from a comic in this way.

panel from Kiyohiko Azuma's Azumanga DaiohThis is the first panel of one of my favourite Azumanga Daioh (あずまんが大王) strips, by Kiyohiko Azuma. On the literal level, we can identify all the objects depicted: the face of Osaka (one of the main characters), her hair and part of her neck, and a bed with its various components. We are also able to tell the flower pattern on the blanket apart from ‘real’ flowers-as-objects. I think we’re still within the realm of the denoted message when we say: this is a sleeping girl.

Before we tackle the symbolic message (even though literal and symbolic message are perceived simultaneously, says Barthes), let’s look at the linguistic message. There is no writing in this panel, but in this case, Barthes is right when he says there’s always some writing nearby. Consider the complete four-panel strip (from the German edition published by Tokyopop):

strip from Azumanga Daioh by Kiyohiko AzumaThere is a heading above the panel and a speech bubble in the fourth panel. The heading reads “Osaka’s New Year’s dream” and tells us that rather than just watching Osaka sleep, we are about to learn what she’s dreaming of. The speech bubble says, “Chiyo, how come you’re able to fly?” Now we need to recall the knowledge of writing that is located even further away from our panel: in previous episodes, Osaka already had weird ideas about the character Chiyo and her plaits. With this knowledge in mind, we can almost guess Osaka’s dream (and indeed, in the next strip we see how Chiyo’s flapping bunches enable her to fly in Osaka’s dream). I’d say this is an example of anchorage: without the writing, the meaning of the image would be unclear.

As for the symbolic message, we need to take into account once more previous episodes, which established Osaka as a slow character who often has wondrous or naive thoughts. This strip is not funny because of the absurd idea of a flying girl – after all, such illogical ideas are common in everyone’s dreams – but because it’s such a typical thing for Osaka to think, no matter whether she’s asleep or awake. The symbolic message may well be to invoke Osaka’s characteristics in order to convey humour. In the last panel, the invoked characteristic are her weird ideas, but in the first panel, which is repeated twice almost unchanged, it’s probably Osaka’s slowness. When in a later episode the livelier character Tomo is shown dreaming her own “New Year’s dream”, one panel of her sleeping in bed is enough to indicate that the following panels show her dream.

So that was, if I got it right, a basic application of Barthes’s theory to a comic panel. A more interesting example would be a panel in which literal message and symbolic message are at odds with each other, or in which the linguistic message acts as relay rather than anchorage. I’m sure there are plenty of such examples once one starts looking.

¹ I’m using Stephen Heath’s translation, “Rhetoric of the Image”, in: Roland Barthes, Image – Music – Text, New York 1977, pp. 32-51.
² Comics Studies have later developed more elaborate systems of the relationship between scriptorial and pictorial elements in comics. See e.g. Nathalie Mälzer’s recent conference paper.

Review: When Manga Came to America

"When Manga Came to America" book coverTitle: When Manga Came to America. Super-Hero Revisionism in Mai, The Psychic Girl
Language: English
Author: Julian Darius
Year: 2014
Publisher: Sequart
Pages: 69
Price: $6.99 (US)
ISBN: 978-1-9405-8903-9

When I first heard about this book, I was looking forward to it (even though most of it had been published before as a series of blog posts), because it seemed to be about roughly the same topic as my PhD research project, or at least some aspects of it. The more other people write about it, the less I still need to write. However, When Manga Came to America turned out to be of not very much use to me, and I doubt it’s of much use to scholarly research in general.

Darius’s book is divided into three chapters, plus an Introduction (pp. 1-11). The very first sentence of this introduction is already telling: “In early 1987, Eclipse introduced American comics readers to manga with three translated series” (p. 1). This sentence tells us that the book isn’t really about the manga Mai by Kazuya Kudō and Ryōichi Ikegami per se, but rather about its American edition. Granted, the title of the book is When Manga Came to America, but it is striking that Darius doesn’t say anything at all about the original Japanese publication – even though sometimes he really should, e.g. when he critisises the narrative structure of the comic on p. 7 (which may be better explained by its original serialised magazine publication than by the three ‘acts’ Darius has made up), when he compares it to other comics published in the US at the same time on p. 41 (instead of comparing it to other manga originally created at the same time), when he talks about Mai as an example of an “extended finite” comic book series on p. 49 (thus again ignoring its original publication format), or when he says that a dialogue “simply feels off” on p. 53 (without mentioning that it’s only a translation).

It’s also noteworthy that Darius says “American comics readers” in that first sentence. Rather than considering the reception of Mai among Americans in general, including those who haven’t read comics before, Darius imagines a specific type of reader who is familiar with superhero comic books and who sees Mai in this context. (It should be noted that Darius doesn’t talk about the actual reception history of Mai – i.e. no empirical evidence is presented, such as contemporary reviews or interviews -, and probably not even his personal reception as he was only 10 years old when Mai first came out in May 1987. Instead, he speculates about what the reading experience must have been like at that time.)

The first two chapters must be read with this particular angle in mind. In the chapter “The Depiction of Super-Powers in Mai, the Psychic Girl” (pp. 12-40), Darius offers a lengthy summary of the plot of the series with an emphasis on its supernatural, psychic elements in comparison to superpowers in standard American superhero comic books. While doing so, Darius dishes out his subjective, ungrounded opinions, e.g. “there’s no doubt that the third act is rushed” (p. 30).

The second chapter, “Mai, the Psychic Girl and Revisionism” (pp. 41-49), is much shorter and examines Mai in the context of superhero comics of the late 80s, above all Alan Moore’s Marvelman/Miracleman. Again, I find this comparison somewhat pointless without considering the original publication, as the American readers must have been well aware that Mai was a Japanese comic. Darius makes quite a bold statement again when he says, “Mai was published in English at the height of the revisionism, and it couldn’t help but be understood in this context, even if this wasn’t the context in which Mai had been created in Japan” (p. 41).

In the third chapter, “Sexuality in Mai, the Psychic Girl” (pp. 50-66), Darius thankfully does away with the superhero comparisons. He identifies several scenes with nudity or other sexual content and comes to the conclusion that Mai, even though it “is pretty restrained” (p. 63), is really “a story of sexual self-discovery” (p. 65), and “a story in which this [Mai’s psychic abilities] is linked to sexual maturity” (p. 58). This looks to me as if Darius misses the mark and over-interprets the story in a desperate attempt to make sense of the occasional suggestive imagery (which definitely feels out of place in this comic). The question is indeed “whether Mai is entirely of good taste” (p. 63). This chapter gets more awkward the further Darius digresses from the comic, saying things like “breasts, in particular, are a traditional symbol of nature’s bounty” (p. 61), or “healthy male brains often find girls attractive at around the age that girls become capable of reproduction” (p. 62). (Why male, by the way? What about female readers?) This excursus on paedophilia, by the way, contains the only footnote reference in the whole book. I’m not saying a text without formal literature references can’t have any scholarly merit, but it is certainly indicative.

It’s not as if When Manga Came to America wasn’t worth your time or your money, as it is both short and affordable. The reason why I’m reviewing it at all is connected to a larger issue within Comics Studies, as well as within Humanities as a whole: there is a widespread attitude that a researcher may choose which secondary literature to read and cite, and which to ignore – unlike in the Sciences, where it is usually clear which literature needs to be cited in a text on a specific topic. I’ve always felt the Humanities could and should do better in this regard. Therefore, I’d like to suggest how to deal with Darius’s book in a scholarly context, because it is a piece of secondary literature that isn’t easily defined as either a scholarly book or a journalistic review. Its merits notwithstanding, in my opinion, this book may be safely ignored in scholarly discourse, at least as a secondary source alongside proper scholarly texts. Of course, the author probably never intended it as a contribution to scholarly discourse in the first place. And who knows, When Manga Came to America may still serve as an entertaining book for fans who can’t stop thinking about Mai, the Psychic Girl, and who appreciate the opinions of another well-read fan.

Rating: ● ○ ○ ○ ○

Akira Lego

Already in October/November last year, on the occasion of Designer Con in Pasadena, California, some people re-created comic book covers with Lego bricks and put pictures of them online, mostly on Flickr in the group Comic Bricks! and/or using the tag “comicbricks” (Nerdist and several other websites reported). Some of these pictures are fascinating in the way in which a three-dimensional object was extrapolated from a two-dimensional cover, e.g. the iconic “Demon in a Bottle” Iron Man cover.

The most interesting Lego cover from my perspective is, of course, the only one in that Flickr photo pool that is based on a manga cover: Akira #31. Its mere existence among otherwise American comics is remarkable. Then again, an issue from the old, coloured, 38-part Epic Comics edition was used, which was more like a US comic book than the later black-and-white collected volumes.

detail of Akira #31 photograph by Tommy Williamson

The creator of this Lego cover retained the abstract elements of the cover and interpreted them as a sort of window frame through which we see Kaneda and Kai riding the ‘caretaker’ robot. In contrast to the original cover, though, one leg of the robot extends through this frame in the front, while another robot leg can be seen extending to the right behind the frame (and also parts of the wall in the background to both sides), making for an imaginative compromise between 2-D and 3-D elements which, as far as I’ve seen, hasn’t been tried in any of the other Comic Bricks covers.


Get every new post delivered to your Inbox.

Join 71 other followers