Background and purpose of this study

Previous research has shown amply that humorous Internet memes can serve as users’ responses to, and commentaries on, the current socio-political or socio-cultural facts, with COVID-19 being no exception. The COVID-19 pandemic has inspired a plethora of relief-giving (Bischetti et al., 2021) online humour. In turn, this has given rise to a flurry of research that, with very few exceptions (Murru & Vicari, 2021; Dynel, 2021), addresses mainly the recurring topics of memes (Lemish & Elias, 2020; Aslan, 2021; de-Saint-Laurent et al., 2021; Norström & Sarna, 2021). This paper contributes to the COVID-19 humour research, based on an automatically generated corpus. Rather than analyse topics, this contribution answers a twofold research question: it distils the categories of the memes in the light of their form, content and stance (Shifman, 2013) and, most importantly, it examines these categories diachronically. Such a diachronic approach does not seem to have been adopted in meme research so far, while online humour (taken as a whole) has been subject to diachronic analysis only intermittently (Demjén, 2018; Archakis & Tsakona, 2021).

This study zooms in on COVID-19 mask memes, a salient and socially relevant COVID-19 topic (besides quarantine and lockdown), circulated from January 2020 to January 2021, the time which brought the first shocking news, numerous controversies and diverse opinions about mask-wearing (unlike other counter-measures before the arrival of vaccines: social distancing and hand hygiene).

This project is premised on the assumption that the memescape that addresses a current and “hot” topic, such as that of COVID-19 mask memes, will respond swiftly to the changes in socio-cultural and political arenas. Based on the assumption that memes reflect the current events (Ross & Rivers, 2017; Dynel & Poppi, 2021; Norström & Sarna, 2021), the central hypothesis is that COVID-19 mask memes should vary across the months, emerging in the wake of the socio-political developments concerning the pandemic and its effects only to fall into oblivion when the memes’ novelty has worn off, or when the specific issues have become irrelevant. The examination of the automatically generated corpus, in qualitative and quantitative terms, yields interesting conclusions about the circulation of not only COVID-19 mask memes, but also memes in general. First, the definition of an (Internet) meme is fine-tuned to better account for individual items and memetic virality.

Memes

According to the original biology-rooted definition proposed long before the digital age, a “meme” is a cultural unit, a piece of cultural knowledge, that infects individual minds through copying or imitation; thereby, it proliferates among individuals and across societies through replication, potentially undergoing necessary mutation, very much like the “selfish” gene (Dawkins, 1976; Blackmore, 1999; Shifman, 2013; Wiggins & Bowers, 2015). Hence, as Wiggins (2019) summarises, the Dawkinsian meme is an idea that spreads virally (originally outside of the Internet and contemporarily via the Internet) without any discernible modification or remixing. However, this idea of a change/remix is a crucial definitional component of an “Internet meme”, which is how it is distinguished from a viral one. The latter snowballs across online platforms in an unchanged form (Shifman, 2013; Wiggins, 2019) and may serve as the basis for an emergent meme (Shifman, 2013).

Consequently, the key difference between the two notions is variability; “whereas the viral comprises a single cultural unit (such as a video, photo, or joke) that propagates in many copies, an Internet meme is always a collection of texts” (Shifman, 2013: 56). The pending queries are how similar these “texts” (or rather, multimodal, digital items) need to be and how this similarity can manifest itself in order for the items to constitute a meme.

Based on Shifman’s (2013: 7) proposal, the “texts” that form “a meme” are “a group of digital items sharing common characteristics of content, form, and/or stance”, which allows for much leeway in terms of what qualifies as an item belonging in a meme. Thus, Shifman’s (2013) conceptualisation presents a meme as being constituted by numerous digital items taken together, which is often misquoted and misunderstood. This brings the discussion to the problem of the polysemy of the term “meme”, which Shifman’s definition skilfully straddles.

A meta-analysis of the ample communication and socio-cultural research on Internet memes leads to a conclusion that there seem to exist two markedly different approaches to this notion, with many works making use of both senses without actually recognising the polysemy. On the one hand, in line with the Dawkinsian conceptualisation, an Internet meme is conceived as a viral idea or a cultural item that gets replicated in various online instantiations necessarily through mutation and remixing (Mielczarek, 2018; Wiggins, 2019). In this sense, memes are thought to be viral (Wiggins, 2019) in line with Dawkins’s thought even if, at the same time, they should not be mistaken for digital items known as “virals”. On the other hand, presumably, a more popular definition (which is in line with its folk understanding) presents an Internet meme as the very instantiation of a viral idea, namely as a multimodal digital item, typically humorous and “circulated, imitated, and/or transformed via the Internet by many users” (Shifman, 2013: 8-41). Overall, memetic productivity and creativity are the epitome of the social media culture labelled “remix” (Lessig, 2008) or “convergence” (Jenkins, 2006). Incidentally, on platforms dedicated to meme circulation, novelty and creativity are the central goals, with users earning their kudos by submitting items that are well received (e.g. awarded or upvoted) by other users.

In order to avoid confusing different understandings of what an “Internet meme”, or just a “meme” (taken by default in studies on online communication), is and conflating virals with memes, a distinction is drawn between a “meme” and a “memetic construct”. A meme is thus understood here as each individual multimodal instantiation of a memetic construct that, in turn, coincides with the Dawkinsian meme, that is the viral idea manifest in different iterations. A memetic construct that echoes across specific memes may concern content, form or stance; it may materialise, for instance, as an ideological meaning or a memetic template endowed with an inherent interpretation to be merged with the message carried by whatever is superimposed on it.

While a memetic construct may take up a stable position on memescape, specific memes typically arise only to “fall into virtual oblivion, outdone by newly minted iterations” (Mielczarek, 2018: 71). Many authors (Knobel & Lankshear, 2007; Shifman, 2013) have noted that memetic constructs are elusive, ephemeral entities showing a transient nature: they appear all of a sudden, generate some interest and quickly become a thing of the past. However, some memes and memetic constructs may resurface with old or new meanings in line with the basic premise of remix culture (Mielczarek, 2018). Moreover, as this study indicates, individual memes may actually go viral, being reposted in the same form across platforms.

Qualitative findings

The corpus, comprised of eight monthly sub-corpora, is divided into ten mutually exclusive categories clustered into four groups centred around: 1. The type of mask worn by a subject in the image (“reports”, “parodies”, “pranks” and “spoofs”), 2. Evaluation of mask-wearing (“corrective”, “chastising” and “skeptical”), 3. Masks being commonplace (“sign of the times” or “pre-COVID 19 times”), and 4. Masks as tools (not the conceptual focus) facilitating memetic humour construction (“props” or “background”).

As shown in Dynel (2021), many COVID-19 mask memes involve subjects, or what Kress and van-Leeuwen (2006) call “represented participants”, that are typically people but also animals shown in the images by “interactive participants”, i.e. the new media users who (re)post the memetic content online. Sometimes, the represented participant may coincide with the interactive participant, as is the case with selfies posted online for the first time, or the two may be collaborators. By contrast, represented and interactive participants may be strangers, the former being oblivious to their becoming the foci of memes, as is the case with the category dubbed “reports” (Figures 1A-1C).

Reports concern the memetic construct of masks and other items of clothing sincerely worn by subjects for protection against COVID-19. These subjects are typically captured in photographs unknowingly in public spaces by observers who find the protective measures so bizarre as to be amusing. These include make-shift items, such as the peculiar composition of plastic (Figure 1A), and other uncanny manifestations of protection, such as the multiple surgical masks worn simultaneously (Figure 1B) or the mask on the pet’s head (Figure 1C). These photographs seem to have been taken at the beginning of the pandemic (as evidenced by the month of the original postings), when informational confusion about the new virus and the dearth of protective measures caused panic-stricken citizens to go to any lengths to avoid the virus. The strangely dressed humans and animals can be said to have unintentionally become the source of amusement, with the humour being the intended outcome of the meme (re)posters’ online practices.

The “reports” category can be formally juxtaposed with memes also centred on subjects with peculiar masks or full attire, albeit worn (or only presented as if worn) in public spaces mainly for the sake of fun rather than protection, such as the newspaper beak (1D) or the dragon’s head (1E). This category comprises two memetic constructs, “pranks” and “spoofs” (Dynel, 2021), which may sometimes be difficult to tell apart in practice, given users’ Photoshop skills and lack of contextual evidence (such as the passer-by laughing at the subject in 1D). The label “prank” concerns subjects who mean to amuse other people they encounter in public places, whether or not cognisant of the pictures being taken and their imminent meme status. Spoofs, in turn, involve subjects whose looks are purposefully manipulated with the use of digital technologies, for the sake of only online receivers’ fun, with the general idea being to poke fun at a subject’s weird protection that they are (allegedly) wearing. Both pranks and spoofs contain an element of humour-oriented deception concerning the intention behind the mask-wearing and/or the actual act of mask-wearing.

Another related category, mask parodies (Dynel, 2021), is a memetic construct that encompasses images, typically selfies, of subjects at home, exhibiting their newly designed masks, which they clearly have no intention of wearing in public but which are meant to humorously echo creative self-made masks (cf. the reports category). A case in point is the meme in 1F. It reports the unfortunate result of copying the popular advice involving the transformation of a padded bra cup into a mask; the predictable failure is due to the cup size, with the woman’s masks being much too big for her and her children. This category also includes images of subject-less masks, such as those carved from bacon (1G). Overall, the three categories of memes rely on peculiar, often creative, masks being a source of humour, either as intended by the subjects, who may coincide with meme producers (pranks/spoofs and parodies), or without this intention (reports). It is also the concept of an ordinary medical mask per se or an otherwise known type of mask that may be a source of humour, as the following two categories illustrate, both relying on cultural references (Figure 2).

The category labelled “sign of the times” shows a salient memetic construct: masks being superimposed on faces known from various cultural artefacts, whether historical or contemporary, such as works of art (e.g. statues or paintings like Van Gogh’s self-portrait in 2C), publically known individuals (e.g. Grumpy Cat in 2A), or cultural symbols, such as Benjamin Franklin on the US one-hundred-dollar bill (3B). Such memes play with the idea of the prevalence of protective masks in the COVID-19 times. Sometimes, an additional humorous effect arises when special features of the represented person are considered, as is the case with van Gogh’s mutilated ear that would have made it impossible for him to wear a surgeon’s mask.

The related category of pre-COVID-19 masks pertains to memes that feature various kinds of masks worn before the COVID-19 era, whether by fictional characters (Hannibal Lecter in 2D) or famous people (Michael Jackson in 2E). This memetic construct indicates that the topic of COVID-19 masks evokes users’ memories of pop-culture masks, whether or not used for anti-viral protection. The protective function of masks is the topic of three other meme categories (Figure 3), two of which endorse adequate mask-wearing (3A-3D), whilst the third one questions the need for mask-wearing and its efficiency (3E-3F), collectively reflecting the long-standing medical, political and legal debates held both on the personal level and in public media. In these memes, it is the topic and stance that determine the memetic construct at hand.

“Corrective” memes present medical rationale supporting the role of masks in limiting the spreading of the virus and offer advice on how masks should be worn properly. This is done, for instance, through the multimodal simile (3A) or the visual representation of the adequate vs inadequate (and sometimes even absurd) mask-wearing practices with the help of a dog (3B). In this vein, the “chastising” category encompasses memes that express a negative evaluation of people wearing face masks wrongly (as in 3C, below the nose, as done by Sheila Jackson Lee, a US politician presented by the meme creator as holding a special function during the Coronavirus crisis in the USA), or wearing masks that cannot possibly guarantee protection (cf. 3D, which shows a photograph of a television screen with a brass orchestra wearing masks with holes allowing the musicians to play but making the masks useless as protective measures). By contrast, memes classified as “skeptical” criticise masks based on the assumption that they do not stop the virus from spreading, as represented through the multimodal metaphor in 3E, as well as indicate some people’s imperviousness to the repeated requests that masks be worn for the sake of mutual safety (cf. 3F, which merges a cynical statement with a metaphorical representation of a relaxed attitude).

Overall, the three categories of memes take mask-wearing as the central topic, with users presenting their stances on this issue. These have been isolated from among the memes constituting an eclectic group that relies on masks but does not address the topic of mask-wearing as the central focus and (unlike all the other previous ones) does not evince one memetic construct in the form of a single topic. This group contains a category dubbed “props”, which is juxtaposed with the one that merely involves the use of masks as the “background” (Figure 4).

The “props” category is very broad and seems to be closest to memes understood as standard multimodal jokes, whilst the other categories discerned so far appear to be specific to COVID-19 mask memes. This heterogeneous category cannot be broken into any neat subtypes. However, some of the memes that use masks as their joke props can be thought of as humour based on wordplay (cf. the Corona beer inspiring fear in other drinks in the fridge hiding behind a mask in 4A) and visual play (the exotic-holiday beach image involving the use of surgical masks as waves in 4B) without much informative import or stance-taking, whilst referring to COVID-19 masks in one way or another. The point of these memes is just to make light of masks, which serve as joke props in the memes. Other memes in this wide-ranging category take masks as their point of departure to make personal commentaries, as is the case with the meme in 4C. It features a pop-cultural, intertextual reference to Bane (Batman’s masked opponent from “The Dark Knight Rises”) with a quotation from the film (top) followed by a comment on the COVID-19 reality. Similar to this item, many other memes address the real or potential consequences of people having their faces covered, such as being secretly mean to others with no repercussions (4D) and getting a peculiar sun-tan (4E).

By contrast, albeit tagged “mask memes”, some items seem to involve masks only as secondary items helping form the background of memes, rather than constituting their pivots. Such is the case with the plague doctor’s mask accompanying the text that addresses the history of plagues occurring every 100 years in 4F. Moreover, memes belonging to this category may not be primarily, or only, about COVID-19 per se, as is the case with the critical meme in 4G that metaphorically juxtaposes unwise mothers’ (irrational) means of coping with the coronavirus and lice at schools: an angry rebellion with no protective measures applied (represented by a bare-handed fight scene from “Fight Club”) and a counter-action taken in protective gear from head to toe, respectively.

Quantitative and diachronic findings

The sizes of the monthly corpora of individual memes, with no duplicates included within each month (n=1,008), are indicative of the interest a given category (except for props and background coinciding with a memetic construct) generated among meme posters in the specific months (Figure 6). The colours represent the groups of memes: 1. The type of mask worn by a subject in the image (reds), 2. Evaluation of mask-wearing (greens), 3. Masks being commonplace (blues), and 4. Masks as tools in memetic humour construction (violets).

The general quantitative findings emerging from the collected corpus are hardly surprising. As predicted, the most memetically productive month in the corpus is April 2020 (n=203), the month of international lockdown, which is reported to have significantly affected the use of digital media on the whole (Nguyen et al., 2020). The stay-at-home time must have promoted users’ social-media memetic productivity concerning the COVID-19 pandemic. By contrast, the smallest number of posts in July (n=99), together with the rather low number in May (n=112), is a natural consequence of nations coming out of lockdown and the holiday season in many countries, when most restrictions were lifted, potentially causing people to spend less time engaged in memetic production online. Therefore, all the categories were in a slump in May and July only to bounce back in September. The second least amply represented month is January 2020 (n=102), the month when the information about the new coronavirus only started pouring in, and the topic could not yet have been of top priority.

It had been expected that “reports” would be the category showing the highest frequency of occurrence in January 2020, which saw the first reports about the virus in China and, soon afterwards, in other Asian countries, as well as about the dearth of medical masks there, with panic caused by the lack of precise medical instruction. At the same time, there was no imminent, palpable threat to the entire globe blocking the humour experience at the expense of the subjects (Bischetti et al., 2021 on the spatial and temporal distance from a tragedy as a factor in the experience of humour).

However, while the “reports” category did achieve its peak in January (n=24), it also had significant representation in March (n=13), April (n=21), September (n=15) and November (n=15). Surprisingly, reports were salient in later months too, often in the form of reposts, potentially testifying to the perceived humorousness of the images showing the desperate people and their self-made protection (on themselves or even their animals) to be made fun of, even at the time when the grave COVID-19 consequences were recognised worldwide.

On the other hand, the categories of parodies, pranks and spoofs enjoyed their greatest popularity in the months following January 2020 and the reports from Asia. Each achieved its peak in April (parodies: n=43, pranks and spoofs: n=15), which is the indication of people wanting to have fun under lockdown. Taken together, parodies (n=174) outnumber pranks and spoofs (n=55), which may have to do with the ease of production.

Presumably, it takes either more courage to go out in a peculiar costume or more luck to see and take a picture of a person in peculiar attire; likewise, one may need more technical skills to convincingly manipulate an image than to take a picture of a self-made mask (whether of their own design or merely copying what someone else has already done) or a selfie in a peculiar mask worn at home. Overall, parodies, pranks and spoofs form the second biggest group in the corpus (n=229), and their presence is marked not only in April (n=58) but also in other months, most notably: November (n=36), March (n=32) and September (n=28).

Similar to parodies, pranks and spoofs, the sign-of-the-times and pre-COVID masks were, in general terms, amply represented across the months, starting in March (when masks began to be commonly used worldwide), altogether constituting the third largest group of memes (n=155). The top four months are April (n=40), September (n=27), November (n=22) and March (n=20). These two categories are also a form of memetic play, based on the simple assumption that masks have become commonplace or that they were widely used and known in the past. At the same time, the exceptionally high number of pre-COVID-19 mask memes in April may be an indication that users were on the lookout for the validation and normalisation of masks, however peculiar the masks they were referring to could be (cf. the mask worn by the fictional serial killer Hannibal Lecter).

Insignificant differences across the months aside (corresponding with each monthly corpus size), the corrective and chastising memes (n=106), as well as the opposite category of sceptical memes (n=23), showed almost an even distribution across the months for the best part of the year following January 2020. This category seems to reflect the long-lasting discussions and controversies around the efficiency of masks as protective measures, as well as the salient cases of media personas wearing their masks inappropriately. Based on the data at hand, it can be hypothesised that meme creators tend to trust the efficiency of masks, with sceptics being in the minority.

“Props” is the category that features most prominently across the months (except for April, when it is slightly superseded by parodies), with the lowest (and yet still significant) count (n=23) in May, and the highest result (n=49) in September. With these scores, it is the largest category (n=275) in the corpus, taking into account even grouped categories (cf. the reds, greens and blues in Figure 5). This is hardly surprising since this is a large category that encompasses numerous memetic constructs. Some diachronic variety can be observed across the months, inspired by, and reflecting the social, cultural, and political context, as observed by users. For example, given the early reports about COVID-19 originating in Wuhan, the first few months of the global pandemic saw memes such as the one in 6A based on an advertisement of masks sent (ironically enough) from Wuhan and human-tested there, as the meme author’s highlights indicate.

Additionally, the corpora from January and March (the month the pandemic was announced across the world, following the reports of the virus in China and other Asian countries in January, and Italy in February) contain memes about people being scared of getting infected and memes criticising people spreading the virus by travelling, while memes posted in the later months as the pandemic is raging tend to comment on lockdown, quarantine, and absurd governmental decisions, among other things. Moreover, some memes from the props category start appearing after specific events, as is the case with a public recommendation made by Dr Anthony Fauci, the director of the National Institute of Allergy and Infectious Diseases, about the benefits of double-masking, which the 6B meme creator ridicules through a multimodal hyperbole. However, some memes seem to be topically universal in the COVID-19 world, such as the playful meme in 6C that alludes to the pandemic-induced need of mask-wearing as if it is worse than a bank robbery from a bank employee’s perspective. Needless to say, memetic constructs may feature in many memes, as is manifest in 6D, which centres on a screenshot of news presenting an information notice that contains a typo (“bananas” vs “bandanas”), inviting humorous activities in response, three of which are shown at the bottom of the meme. Interestingly enough, the bottom-right-hand-corner example must be a repost of an independent meme featuring the same information notice in it.

Finally, the presence of the marginally relevant “background” category in the corpus seems to be the consequence of (over-)tagging that users perform in order to boost the findability of their memes. At the same time, the preponderance of such memes testifies to the user-perceived salience and/or normalised prevalence of masks in the socio-cultural world affected by the COVID-19 pandemic. On a general note, the findings indicate that memes and memetic constructs are not as short-lived as might be expected. Apart from the fact that memetic constructs persist over the months under investigation even if their relevance is limited, individual memes tend to be reposted not only within but also across months, sometimes going viral. Within the corpus, a substantial number of items (n=59) occur in more than one monthly subcorpus, and the most recurrent meme (6C) resurfaces four times across the months.