Memes are everywhere. That’s their most striking feature, really : wherever you go on the Internet, you’re bound to find something that more or less resembles it, from an image macro, to a twisted boomer comic or a completely surreal image that you do not fully understand nor wish to. Memes have become somewhat of a symbol for Internet culture, its capacity to make anything go viral and its often absurd, but somewhat universal, humour. This trait, that I would call spreadability, is widely considered to be what makes a meme, a meme. Biologist Richard Dawkins, who coined the term, used it to describe all the repetitive cultural elements and practices that are transmitted from person to person. As another biologist defines them, memes are “stories, songs, habits, skills, inventions and ways of doing things that we copy from person to person by imitation.” (Blackmore, 2000, p.65)
What’s striking is that the same biologist, as early as 2000, used an example taken from the Internet to explain what memes are – what we would now call copypasta :
“The most obvious examples of this phenomenon are “viral” memes. Chain letters (both hardcopy and e-mail) consist of little bits of written information, including a “copy-me” instruction backed up with threats (if you break the chain you will suffer bad luck) or promises (you’ll receive money and you can help your friends). It does not matter that the threats and promises are empty and your effort in copying the letters is wasted. These memes have an internal structure that ensures their own propagation.” (Blackmore, 2000, p.66)
One could argue over whether or not copypasta really are memes, but what interests me in this example is the last sentence : the fact that it is the “internal structure” of the meme that makes their spreadability possible.
This is in fact quite the important statement because it diverts attention from memes as cultural practice (something to be shared, created, etc.) to memes as objects. Many studies that focus on Internet memes specifically (in opposition to the biological concept) have followed the former approach : see, for example, this definition (Davison, 2012, 122) :
“An Internet meme is a piece of culture, typically a joke, which gains influence through online transmission. While not all Internet memes are jokes, comparing them to offline jokes makes it clear what makes Internet memes unique: the speed of their transmission and the fidelity of their form.”
This invites us to focus on what I would call “external factors” of spreadability rather than the memes themselves : how are memes created, shared and understood (or not), and not how memes fundamentally function. For that reason, I’d like here to offer another approach, which would look at the internal factors : that is, how the structure of the memes themselves enables them to go viral. I will show that memes can be construed as settings, that is as a specific manner to organize data (1). I will draw on already existing meme studies material, but mostly from anime and otaku scholar Hiroki Azuma and his main work on these questions, Otaku : Japan’s Database Animals.
This choice is motivated by the fact that Azuma’s insights on anime and otaku are similar to the points I’d like to make in regard with memes ; I believe his views on postmodern media consumption are precious even when one looks outside of anime and otaku-related (otaku-kei) media. Indeed, Azuma’s reasoning goes as follows : otaku-related media are not a specifically Japanese phenomenon, but must be understood in relation to a wider historical movement, that of postmodernity. Operating on such an assumption, Azuma then proceeds to establish what, in anime and its consumption by otaku, is postmodern ; and he does so not by looking primarily at external factors (technology, economical and social conditions, etc.), but by studying anime from a purely formal and structural level : that of character designs and types, situations, etc. This is the same method I’d like to try on memes.
The structure of memes
To my knowledge, there has already been at least one attempt to describe memes from the structural level. I am going to quote his definition in its entirety, then discuss it .
“A meme can be separated into components. I propose three: the manifestation, the behavior, and the ideal.
The manifestation of a meme is its observable, external phenomena. It is the set of objects created by the meme, the records of its existence. It indicates any arrangement of physical particles in time and space that are the direct result of the reality of the meme.
The behavior of a meme is the action taken by an individual in service of the meme. The behavior of the meme creates the manifestation. For instance, if the behavior is photographing a cat and manipulating that photograph with software, the manifestation this creates is the ordered progression of pixels subsequently uploaded to the Internet.
The ideal of a meme is the concept or idea conveyed. The ideal dictates the behavior, which in turn creates the manifestation. If the manifestation is a funny image of a cat and the behavior is using software to make it, then the ideal is something like “cats are funny.”” [Davison, 2012]
I describe such an approach as “structural”, it’s because it divides the memes into “components” that interact with each other and then proceeds to analyze each of them. However, it does not fully match my own requirements. Indeed, Davison still considers spreadability to be the core element of the meme ; for this reason, when discussing the parts of the meme, he in facts discusses three distinct acts. To put it another way, to focus on the meme as something that is spread invites to look at it as something that acts and is acted upon. This is most evident in the case of the behavior : it is an “action”, that of modifying or sharing preexisting data (an image, a text, a video, etc.). In the same way, the idea is something active, because its goal is to provoke a certain reaction : in the instance of “cats are funny”, it is both the proposition itself that is to be shared, and laughter at the sight of funny cats.
But this approach entails another problem. Davison identifies three layers in the meme, the ideal being the most fundamental because it “dictates the behavior, which in turn creates the manifestation”. This is what I would call an intellectualist or idealist point of view, which considers that there is an idea of the meme that exists before and beyond its individual manifestations ; it disregards the material conditions of emergence of the meme, that is the free flow of images and informations on the Net. Moreover, it dictates an approach focused on the meaning rather than the form of the meme. Davison himself acknowledges it when he finds that his characterization of the meme is unable to give a satisfying account of a real meme, the “Advice Dog” :
“The ideal of the Advice Dog meme is harder to describe. The meaning conveyed by any single Advice Dog macro can vary wildly. Some have ironic meanings, while others have aggressive or offensive meanings. The subject can be a dog that gives advice or a child that celebrates success. So we can say that for Advice Dog, the ideal of the meme is not always replicated from instance to instance. With no qualities recognizable from iteration to iteration, it would seem there is no justification for linking them together as part of the same meme. However, what is replicated from instance to instance is the set of formal characteristics. We are able to identify each instance as part of the larger Advice Dog meme because of the similarities in form and regardless of the differences in meaning.” (Davison, 2012, p.130)
As we can see, Davison stumbles when confronted to image macros, which can be considered to be a very simple form of meme ; while his article was published too early for it, one can only imagine the difficulty he would have to describe late 2010’s trends such as surreal memes.
My own definition instead looks at memes as things : it is not dynamic, but structural. I identify 3 elements in the structure of the meme :
1. The interface. This would be the equivalent of Davison’s “manifestation” (2) : it is the actual object or content we see on the screen. However, as I will show later on, its status is quite different as it is not the concrete manifestation of an abstract idea, but the actualization of elements registered into a database.
2. The template. Put in abstract terms, the template is what regulates the display and structure of information on the interface. It is, as I will show, the “identity” of the meme : if we are able to recognize different interfaces as part of a single “Advice Dog” meme, it is not because of a shared meaning, but because the template, that is the structure of the statements is the same.
3. The content. The content is the information that is inputed in a template as to make it an individualized interface. For example, the general structure of image macros dictates that they be composed of a top and bottom text, the latter being a punch line of some sort ; the content is the actual words of the joke. However, it is not the joke itself, as the way one should interpret the content can be influenced by the template.
To make this clear, we can look at two interfaces that share the same content but not the same template. The content is a simple pun : “Need an ark ? I Noah guy”, which plays on the similarity of sounds between “Noah” and “know a”. But the interpretation one should have of the two memes in fig. 1 is quite different. The meme on the left is an example of the “Good Guy Jesus” template ; first, the joke is made funnier by the fact that it relies on a biblical reference, therefore linked to Jesus. But more importantly, the template puts the focus less on the pun than its meaning : Jesus, being the titular “Good Guy”, offers you a service while making a pun in the process. On the contrary, the meme on the right is an example of “Bad Joke Eel” : as the name makes clear, this templates invites the viewer to consider that the content is, by definition, a lame joke – which this one is. To sum things up, the first template invites us to appreciate the joke and its context, whereas the second one is meant to reminds us that the joke is bad : the first one is meant to be funny, not the second one (at least at face-value : the “Bad Joke Eel” template is also a signifier of irony).
We can then understand that what is most important is not the content or meaning (Davison’s “ideal”) of the meme but its template, as it is both what dictates the interpretation of each interface, and therefore its meaning, and what is actually shared and modified. As it name implies, the template is just a neutral slate that needs to be inputed content. But, as I have just shown, we should be careful with the “neutrality” of the template, as it is in fact it that is the most meaningful in a very literal sense. To give an analogy, the template is just like syntax : you can play with it to a certain extent, but ultimately, it is what determines the structure of your language and how information is displayed. It exists independently from the information (the actual words) and does not determine them ; however, it plays a vital role in the actual meaning of the sentence, which comes in a large part from the presentation of information.
This means that a meme is not just any viral image or content. For example, a random screenshot from a movie, TV show or anime taken out of context and shared is not, in my definition, a meme. Content becomes meme only when the original data becomes a sign, that is loaded with enough meaning to become a template (see Fig. 2).
Memes and anime as simulacra
Now that we have reached a satisfying enough account of the structure of memes, we can see what their parallels are with anime and otaku-related media. Just like we do with memes, Hiroki Azuma  wonders at the spreadability of anime as media. While he does give an account of how anime is spread, he starts with an ontological statement that situate anime at the core of postmodernity. Indeed, he first focuses not on anime themselves, but on their capacity to generate, apparently endlessly, derivative works such as fan comics, video games, etc. This is when he introduces a central concept of postmodern theory, that of simulacrum :
“The prominence of derivative works is considered a postmodern characteristic because the high value otaku place on such products is extremely close to the future of the culture industry as envisioned by French sociologist Jean Baudrillard. Baudrillard predicts that in postmodern society the distinction between original products and commodities and their copies weaken, while an interim form called the simulacrum, which is neither original nor copy, becomes dominant. The discernment of value by the otaku, who consume the original and the parody with equal vigor, certainly seems to move at the level of simulacra where there are no originals and no copies.” [Azuma, 2009, pp.25-26]
The consequences of such a definitions can be appreciated both in the case of otaku media and memes ; I will now proceed to demonstrate the parallelism.
Derivative works and rhizomatic structure. As the above quote shows, otaku consumers render irrelevant the distinction between the original and the copy. But this distinction becomes even more useless when one looks at memes : in memes, the original, that is both the first recorded interface and the origin of the template are useless in understanding the meme. Moreover, just like in anime, the disappearance of the difference lies not only in the eye of the beholder : it is characteristic of the media itself. Vast franchises such as Gundam, Evangelion or Fate where there are as many alternate versions and parallel universes as there are original works, put into question the very concept of “canon”, that is of a legitimate version that should be used as a reference to decode the other entries in the franchise. In the same manner, the very nature of the template is to be open to different kinds of content, to the point where the content itself might become a template (see fig.3). The reference Azuma convokes is then French philosopher Gilles Deleuze and the rhizomatic model, where there is no “inner layer” of deep meaning that informs the surface of media, but “in which signs are linked in diverse patterns over the outer layer alone.” [Azuma, 2009, 31] In other words, there exist only decentralized webs of surface-layer meanings, and no hierarchical organization of statements.
Moreover, memes have increasingly become more “derivative” with the rise of irony as a more and more frequent register of humor. More specifically, memes referencing other memes, or the repetition of template structures evoke the repetitive and reference-based creation of memes.
Dissolution of the grand narrative. Deleuze contrasts his rhizomatic, decentralized model with the tree model, where the “canon” meaning is comparable to the trunk of the tree, and all the secondary interpretations are like branches feeding off it. This can also be understood with the terms of another postmodern theorist, Jean-François Lyotard, who spoke of “grand” and “small” narratives. The grand narratives are those that provide a totalizing interpretations of events : for example, that human history is driven by progress. In anime, according to Azuma, this translates to narrative-driven works : shows like Mobile Suit Gundam or Legend of the Galactic Heroes, which mimic the approach of a historical documentary, are good examples. But in the course of the 80’s and 90’s, according to Azuma, the function and reception of such works changed ; the main catalyst was the series Neon Genesis Evangelion : whereas Evangelion’s convoluted plot and global conspiracies would tend to make it similar to previous works like Gundam, most of these plot elements are, in the end, not resolved and the viewer is left without answers. What is presented in Evangelion is therefore not, according to Azuma, a narrative, but instead a “nonnarrative” : “Many consumers of Evangelion neither appreciate a complete anime as a work (in the traditional mode of consumption) nor consume a worldview in the background as in Gundam (in narrative consumption) : from the beginning they need only nonnarratives or information.” [Azuma, 2009, p.38]
How, then, does this relate to memes ? By the simple fact that the very nature of the meme is contradictory to that of a narrative. This can be considered in two ways. First, memes are often short, whether in content (there is often little text in text-based memes, and where there is a lot, it is often ironical) or in duration (in the case of video or sound-based memes) : this brevity is contradictory with the necessary development of any narrative. Second, there is the fact that, to be spreadable, the content in a meme has to be relatable to a various set of people. This can work in two directions : either the meme targets a specific community with common references, and therefore does not need to present them in a narrative-like structure ; either it aims at no audience and then has be general enough to reach all kinds of people : the experiences shared then have to be basic and obvious enough for anyone to understand them. For example, one could argue that rage comics represented an early, narrative-based type of memes ; but then we should note that 1) the stories of rage comics are precisely the kind of basic experience anyone could relate to that I just described and that 2) rage comics have slowly disappeared, giving way to non-narrative types of memes.
Irrelevance of authorship. The generality of memes entails their anonymity : because the experience related in the meme is one that anyone could have, the actual author of the meme, or subject of the experience, does not matter. Moreover, the spreadability of the memes, and the absence of any “original”, means that whoever created said original is quickly forgotten. The same movement might be seen in anime, with the same causes, that is the multiplication of often anonymous derivative works.
Is there a meme database ?
All of the characteristics I’ve listed are still purely descriptive ; they do not enable to go really in-depth in the nature of the object analyzed or the way it functions. In other words, characterizing memes as simulacra may seem like a necessary step, but it hasn’t taught us anything new. This is why we must ask, just like Azuma, two questions :
“1. In postmodernity, as the distinction between an original and a copy is extinguished, simulacra increase. If this is valid, then how do they increase ? […] In postmodernity, what is the reason for the birth of the simulacra ?
2. In postmodernity grand narratives are dysfunctional ; “god” and “society”, too, must be fabricated from junk subculture. If this is correct, how will human beings live in the world ? […] What becomes of the humanity of human beings ?” [Azuma, 2009, p.28]
To answer the first of those two questions, Azuma introduces his famous concept of database consumption that relies on what he calls a “double-layer structure” [p.33]. Azuma uses the Internet as a model : there is a set of accumulated data, which is then organized to create Web pages. However, in this model, the deep layer of data does not represent a place where one might find deeper meaning, because in it, the data is not organized and only takes shape in the surface of the individual Web pages : “the surface reveals different expressions at those numerous moments of “reading them up”” [p.32] What distinguishes otaku from other users/consumers, then, is the fact that they have knowledge of, and sometimes access to, the database itself. Indeed, Azuma goes so far as to say that “In otaku culture […], products have no independent value ; they are judged by the quality of the database in the background.” [p.33]
This double-layer model can easily be applied to the structure of the meme as I have described it. The level of the database would correspond to that of the templates : just as Azuma assumes the “prior existence” [p.33] of a database comprised of situations, character types, etc. which informs every otaku production, we can postulate that the Internet functions as a database that constantly generates new templates. Websites such as Know Your Meme, which provide search functions and a literal database of existing and past memes, corroborate such an interpretation. This does not mean, however, that the database is a closed one, as if there were to be a moment when one could read it all up. On the contrary, the fact that 1) memes are simulacra and 2) templates do not exist by themselves, but only informed by content, means that the database is constantly fed new memes.
The other, surface layer would then be that of the interface. This means that, if we follow Azuma, interfaces are not simple objects, but the result of the collage of diverse elements of the database. To give a proper account of it, I believe that my third concept of “content” provides the necessary final element that enables us to perfect Azuma’s model. Indeed, while Azuma does explain where the simulacra come from (the database), he does not really focus on how they come into being or are produced in concrete terms. In other words, Azuma lacks a third term that would bridge the gap between the database and the individual works. In the case of memes, this term would be the content : it is the dynamical and individual element that gives the template real existence as it transforms it and gives birth to an interface. With this description, we can understand why I have chosen the word “interface” to describe the individual meme : it is not just a “record” of a meme’s existence, like Davison’s manifestation, but an entrypoint in the database of templates. It is therefore an interactive object.
To show how the meme database functions in concrete terms, I will now proceed to an analysis of a certain kind of memes : ironical and meta memes, which function by referencing other memes or templates. The way ironical memes function is pretty simple, and exemplified quite well in fig. 4, itself a meme : once a meme becomes mainstream, it loses much of its value, which rises in part from it novelty, and either disappears, continues living but takes an ironical meaning, or becomes something else entirely (as the case of the “Doge” meme exemplifies). The way irony works obviously needs something akin to a database : to understand a meme, one needs to know that it is ironical ; to know that it is ironical, one must already be familiar with previous, non-ironical, instances of the same meme.
Meta memes are more complex, as they operate on a higher level and require greater knowledge of the workings of the database. The example in fig. 5 is quite telling : for the meme to work requires those who read to identify the original template just with the dialogue and structure. Furthermore, the meme was already operating on a meta level, as it was originally posted on a Bionicle shitposting Facebook group, but had nothing to do with Bionicle… which was precisely the joke. The accumulation of levels of meta humor or irony are analyzed in similar terms by Azuma, as he takes for example the series Neon Genesis Evangelion, and especially a scene in episode 26, where the characters of the show are suddenly transported in a romantic comedy setting quite different from the original apocalyptic one. Azuma comments on the changes to the character of Rei Ayanami :
“In that parallel world with a completely different history, an Ayanami Rei dwells with a completely different personality. But in fact the scene depicted there was already a parody of an image that had been widely circulated as a derivative work at the time of the original broadcast. That is to say, an extremely warped relationship is interwoven into this work, where the original simulates in advance the simulacra.” [p.38]
But what Azuma fails to note is that the “original”, that is the Evangelion TV show, was already a parody : see, for example, the scene where male protagonist Shinji stumbles onto Rei naked, coming out of the shower, only for it to subvert expectations as Rei does not shriek and slap Shinji as database settings would have it. Therefore, episode 26 is not just a parody of a parody ; it is the parody of a parody of a parody… The same level of complexity is often reached in highly ironical or absurdist memes, which demand a very particular stance from the consumer.
Memes as moe
This position, that invites one to watch media from the point of view of the database, I would qualify it as moe. But before explaining what I mean by that, it is necessary to explain how Azuma deals with this complex idea.
I will not delve here into the origins or etymology of the term, as I believe that it does not teach us much ; however, to understand moe, one needs to comprehend that it has a strange role, as it is neither in the subject, nor the object, but in both at the same time. It’s not in the subject, that is the human consumer, because moe is not just a feeling of love or lust or whatever else : it something that you notice in media. But on the other hand, it’s not just in the object, that is media (an anime, a character, etc.) because moe is as well a particular object (e.g. cat ears) as the affective response it is supposed to trigger. In other words, when I say “cat ears are moe”, I am speaking both about an objective quality of cat ears and the feeling I subjectively feel when confronted to cat ears. This, as I will show, is key to understand Azuma’s account of moe.
The key reason why Azuma focuses on moe is that it is the perfect exemplification of database consumption. Indeed, moe elements (that is, the objects or images that trigger moe as a feeling) can readily be listed to form a database of physical characteristics (big eyes, types of hair, animal ears, paws, etc.), character types (tsundere, kuudere, etc.), clothing (maid costumes, sailor fuku, etc.) and so on. Indeed, according to Azuma, from the 90’s onwards, what became most important in otaku-related media was less the convoluted and/or realistic plots, but rather the combination of moe elements and the affective response they trigger ; in such a context, “each character is merely a simulacrum, derived from the database of moe-elements.” [p.53] As we can see, the data in the database is not just any kind of information : it is comprised of moe elements.
I said earlier that, in the case of memes, the database is made of templates ; this means that the templates are the equivalents of the moe elements, whereas each interface would correspond to the anime character as an addition of moe elements. This analogy holds up because, as I’ve shown, templates dictate the affective response one should have towards a meme : in this regard, they act just like moe.
The fact that moe corresponds to a certain kind of stance towards media, often ironical in the case of memes, can be exemplified by surreal and absurd memes ; the case of fig.6 gives a good, self-conscious example of it : there is nothing funny about the image of a frog and the caption “manslaughter”, but coloring it cyan functions as an indicator that this is a meme (that is, serves the function of template) and is meant to be taken as a joke. And I believe that some, at least, do find it funny. The fact that the template functions as an indicator can be reformulated with Azuma in the terms of postmodern theory, that combine linguistics and economics : the moe element, or template, “is not a simple fetish object [as psychoanalysts of otaku tend to think], but a sign that emerged through market principles.” [p.43] While I have neither the time nor the expertise to demonstrate it, I believe that examining the history and relationships of templates in terms of market principles and competition, would bring a very interesting perspective.
Finally, Azuma proceeds to define moe on an anthropological level, that I believe is still very apt for memes. He argues that the proliferation of moe is the sign of an “animalization” of culture and human beings : that is, a situation where one does not consume because of biological needs, but because he is inside a mechanical cycle of affect and response. In animalized culture, one what seeks from media is not some deeper meaning or lessons about life, but just the satisfaction of a very specific demand : “demanding the right formula of moe-elements that more effectively realizes emotional satisfaction, [the otaku] consume and cull new works one after another.” [p.88]
While I believe that Azuma goes a bit too far in his characterization, and that meme culture is not as animalized as the otaku he depicts, his description does make some points. Most notably when one looks at memes from the perspective of humor. Most scholars and, I believe, most people that appreciate memes, would say that they are, on some level, jokes. That is, memes seek to be funny. However, unlike traditional jokes, memes seldom provoke real laughter (3) ; and as I have shown with the case of meta and surreal memes, they often rely on particularly complex mechanisms. This movement corresponds, in my opinion, to animalization : when consuming memes, one does not necessarily looks for well-constructed jokes that will make one laugh ; one seeks a peculiar kind of enjoyment, that does not necessarily express itself outwards and that can be described as understanding the structure of the meme, that is, in Azuma’s words, consuming the database :
“Within the consumer behavior of feeling moe for a specific character, along with the blind obsession, there is hidden a peculiarly cool, detached dimension – one that takes apart the object into moe-elements and objectifies them within a database. […] suffice it to say that chara-moe cannot be explained merely as fanatical consumer behavior”. [p.53]
To further demonstrate my point and close this essay, I would like to focus on a peculiar kind of memes, those specific to a given community. As I will show, those function on a double level, what Azuma calls the double-layer structure : on the one hand, there is the data, made of references that can only be understood by the community in which it is shared. See, for example, fig. 7 : to get the joke, one has to know the plot of the anime evoked, Mobile Suit Zeta Gundam. But even if one doesn’t, the content obviously refers to some sort of saviour complex that anyone can understand. This level of interpretation corresponds to the outer layer of consumption, that is the affective part of moe wherein the otaku believes he shares a strong feeling for said character. However, on the other hand, there is the other layer, that of the database, that is the template : and to understand it, one does not have to be familiar with Zeta Gundam. Here, the non-fan will not laugh like the fan would ; but that is not necessary, because that is not the goal of the meme. What one needs is just to understand the structure of the joke, to see how the template works ; in other words, the data itself is irrelevant in the enjoyment of the meme : one just has to navigate in the database.
I have not credited the authors of all the memes used here, but they have my thanks, mostly the members of the Facebook group TPM Meme Research and Development : without their ideas and memes, this article wouldn’t have been written.
(1) For the use of the word “settings”, I am indebted to Azuma’s translators, J. Abel and S. Kono ; to explain what I mean by the term, I will quote their own note to Azuma’s text [Azuma, 2009, p.132, n.14] : “The Japanese term settei, translated here as “settings”, has a broad range of meanings. According to the Kenkyûsha’s New Japanese-English Dictionary (5th edition), settei means “establishment ; creation ; institution ; fixation”. But the word is used rather differently in the computer world. Since its inclusion into the Japanese-localized versions of the Mac operating system menus in the late eighties, the term has come to take on the added meanings of “configurations”, “preferences”, “settings”, “properties”, “characteristics” and “data”. In the realm of narratology, settei has come to be associated with the settings of a narrative that constitutes a fictional world – including everything from the time period, the place, and character relationships to physical features such as height, weight, and eye and hair color. We have therefore chosen settings to capture the widest possible range of meanings.”
(2) It might in fact not be that much of an equivalent : although it’s not clear from Davison’s account, it seems that by “manifestation”, he means every individual instance of a given meme. For example, the same meme viewed on my computer and my phone would be, for him, two different manifestations ; for me, they are a single interface.
(3) To make my argument clearer, I would say that one laughs at memes just like one laughs in emoticons. For example, “XD” does not mean that one is literally laughing ; it just serves to indicate that one registers and indicates the fact that the information he has received (or sent) is meant to be taken comically
Azuma, H. (translation J. Abel and S. Kono, 2009) Otaku. Japan’s Database Animals. University of Minnesota Press.
Blackmore S. (2000), The power of memes. Scientific American , Vol. 283, No. 4 (October 2000), pp. 64-73.
Davison P. (2012), The language of Internet Memes, in The Social Media Reader (ed. M. Mandiberg), New York University Press.
Zenner E. and Geeraerts D. (2018). One does not simply process memes : Image macros as multimodal constructions, in Cultures and Traditions of Wordplay and Wordplay Research (ed. E. Winter-Froemel and V. Thaler). De Gruyter.