2.1 Background: Genetics, memetics, or semiotics?

The notion of the meme emerged as an attempt to draw an analogy between cultural studies and Darwinian evolutionary biology (Dawkins, Reference Dawkins1976). Richard Dawkins originally suggested that memes – i.e. trends, catchphrases, social norms, beliefs, and so on – are cultural material that transfer across generations and among populations (the comparison being that memes are encoded units of information which evolve and are societally transmitted by way of repetition and mimicry). The analogy, namely how it reduces the agency of the creators and spreaders of memes to subjects of automated transmission and evolution, has remained widely criticized by cultural anthropologists, biologists, and semioticians, among others (see Fomin, Reference Fomin2019: 331). Memes do not spread themselves, nor do ‘brains’ spread them indiscriminately, as memetics would suggest. Instead, socially-aware agents decide to create, compare, select, package and circulate them (cf. ‘hosting’ a meme ‘infra-psychically’, cf. Castelfranchi, Reference Castelfranchi2001: 21).

Taking consideration from all arenas with stake in the study of memes, Shifman's (Reference Shifman2014: 14) three-part criteria of internet memes, as a subject of cultural and media studies in their own right, are a starting point for understanding text-memes: memes are ‘(a) a group of digital items sharing common characteristics of content, form, and/or stance, which (b) were created with awareness of each other, and (c) were circulated, imitated, and/or transformed via the Internet by many users.’ Of course, when we talk specifically about internet memes, which the field of memetics precedes, the process of this exact diffusion is reinforced and amplified by the affordances of social media (Schmid, Reference Schmid2020: 181, 314).

Literature in the field of linguistics seems to focus on the ‘image macro’ type of meme. Image macros are composed of viral images that are de- or re-contextualized by help of superimposed or adjacent linguistic elements (Bülow, Reference Bülow, Bülow, Koch, Krieg–Holz and Trost2023: 260). They have been studied in relation to their form-meaning correspondence, as pragmatic significations, and as realizations of collective conventions entrenched in (some) speakers’ minds (cf. Shifman, Reference Shifman2014; Bülow, Merten & Johann, Reference Bülow, Merten and Johann2018; Zenner & Geeraerts, Reference Zenner, Geeraerts, Winter–Froemel and Thaler2018; Denisova, Reference Denisova2019; Wiggins, Reference Wiggins2019; Zappavigna, Reference Zappavigna, Stöckl, Caple and Pflaeging2020; Bülow, Reference Bülow, Bülow, Koch, Krieg–Holz and Trost2023). The way in which image macros package a stable image with variable or semi-variable text (e.g. one does not simply X, Bülow et al., Reference Bülow, Merten and Johann2018: 21) is similar to the packaging of a stable text with a variable or semi-variable text in the case of text-memes (e.g. me, an empath, VERB-ing X). The variable text portion (e.g. avoiding people in emotional distress because I'm barely keeping it together myself or noticing everyone at the funeral is crying and sad in examples 1 and 2) is referred to here as the co-text. ‘Co-text’ in this case draws from Schmid's (Reference Schmid2020: 19) notion, as the “utterances preceding and following a given utterance or part thereof.”

The closest formal description of text-memes has been elsewhere assigned the term ‘snowclones’ or ‘phrasal templates’ (cf. Pullum & Whitman, Reference Pullum and Whitman2004; Know Your Meme, 2007–). So-called snowclones would be template-like phrases such as ‘X is the new Y’ or ‘to X or not to X’, for example. The term ‘phrasal template’ is also quite general; it could refer to really any phrase which includes variable slots, such as those above, or anything such as ‘X(adj) as a Y(animal)’ as in ‘blind as a bat’ or ‘stubborn as a mule’, which we argue are not the same kind of linguistic patterns or phrasal forms as text-memes. ‘Phrasal templates’ as well as ‘snowclones’ differ from text-memes with regard to both forms and functions, which is noticeable in how typically simpler, less abundant in text, less narrative-driven, less obscure, and more constrained snowclones and phrasal templates are in comparison to text-memes. In addition, text-memes are products of social media creation and spread. The original purpose and connotation attributed to ‘snowclones’ are also at odds with our notion of text-memes. In an entry to Language Log, Pullum and Whitman (Reference Pullum and Whitman2004) coined the term as a way of describing ‘cliché frames for lazy journalists’. Whereas snowclone emphasizes the imitative, ‘cloned’ property of meaning-making, text-meme emphasizes the creative, ‘text’ property of meaning-making.

2.2 Definition and form: Meme + co-text = text-meme package

Text-memes can be characterized as bipartite units, consisting of a somewhat fixed and a somewhat variable part (see Figure 1), despite that meme-users may likely grasp the entire construction holistically. Otherwise, i.e. if they are interpreted compositionally, it is unlikely that language users are able to recognize them as having a ‘special significance’ such as in their pragmatics or idiomatic-like expression. In the text-meme me, an empath, VERB-ing X, we would identify the fixed meme portion me an empath and the co-text VERB-ing X.

Figure 1. Visualization of the bipartite text-meme as ‘meme’ plus ‘co-text’

Memes and their co-text are complementary. Co-texts are the core of a speaker's communicative point in terms of propositional content. Memes in turn make a strong contribution to perspectivizing this content, e.g. by obscuring, enriching or questioning it and, most importantly, by placing it in the interpersonal relationship between writer/speaker and reader/hearer. Memes cushion the co-text, which also means that certain co-texts call for certain memes.

Co-texts are mostly variable, in that they can take on the topic of really anything. But in order for text-memes to cohere and have a maximal effect and reception as text-memes, co-texts must stick to formal and semantic conventions imposed by the meme. For example, it would be unconventional if the aforementioned text-memes were to instead have the following co-texts:

These two fabricated reconstructions of examples 1 and 2 illustrate how co-texts are only variable to an extent. Example 3 deviates from the convention imposed by the meme, in that me, an empath is followed by a relative clause modifying me. One would usually expect the meme to be followed by a verb phrase where the head verb takes the progressive ing-form. A particular semantic type of verb is also likely expected, typically either a psych verb such as sensing, seeing, or feeling, or at least some verb related to a scenario of having feelings akin to empathy (see avoiding in example 1). Although formally typical, example 4 is unconventional regarding the semantics of attending. There is no direct or obvious relation between (attending an event) → (being an empath), but maybe only when (attending a particular event) + (perceiving it in a particular way) → (being an empath). We could however hypothesize that speakers give less effort to ticking certain grammatical and even semantic boxes of the co-text than they give to reporting a meme somewhat uniformly. Again this expectation is based on the assumption that some degree of holistic processing is required for successful understanding of text-memes.

The meme portion itself is formally constrained as well, but also flexible to some degree. Speakers have leeway in how they report the meme portion. As is also the case with co-texts, limitations to this creativity are imposed by recognizability. After all, memes have to be repeated uniformly enough to be recognized as a meme in the first place. One often finds ambiguous meme usage though:

In examples 5 and 6, speakers identify themselves as empaths given some explanation, e.g., feeling upset by death, or having a particularly difficult week. One question here may be if other meme users would analyze these examples, despite variation in the stable meme portion, as evoking the me, an empath VERB-ing X text-meme. It could also be that empath is a current buzzword, and the phenomenon of calling oneself an empath is on-trend, apart from the text-meme construction. Such examples might be considered highly modified allusions to the meme, which rest on the key notion of empath and references made to the speaker. As an empath, when X, then Y (example 5) and the empath in me VERBs (example 6) deviate quite a bit formally from me, an empath, VERB-ing X. The question lies in the flexibility of readers to accept the meme in various altered states: e.g., if omitting commas, as in me an empath VERB-ing, is OK, and other changes in the noun phrase, as in me, as the empath I am, VERB-ing, are also OK, but I am an empath because X is not OK?

The issue of variability is probably quite functionalized, however, meaning that deviations from the norm would be recognized and more likely to be found acceptable by the ‘meme-literate,’ but not necessarily by the meme-non-literate (Procházka, Reference Procházka2014; Merten, Bülow & Johann, Reference Merten, Bülow and Johannforthcoming). We can therefore perhaps settle on understanding text-memes as formally ambiguous; as forms that must nod to both to variability and stability. Yet to deviate from the norm and still be recognized as a text-meme requires that the particular text-meme has gained enough traction as a conventionalized ‘utterance type’ (Schmid, Reference Schmid2020: 19–28). Speakers have to know the formal, semantic, and pragmatic conventions of co-texts and memes, which are established and reinforced by usage and familiarity with that particular text-meme. And in order to become ‘proficient’ in text-meme usage, speakers must internalize specific formal and contextual characteristics of not just that one instance of a text-meme, but of an entire cache of other ones.

3.1 Phatic, indexical/interpersonal, intertextual/poetic, and social functions

Initially our interest in memes and their communicative functions was sparked by the suggestion that memes mainly, or even only, communicate phatically, among an internet with increasingly ‘empty’ media (cf. Miller, Reference Miller2008; Varis & Blommaert, Reference Varis and Blommaert2015). Jakobson's (Reference Jakobson and Sebeok1960) phatic function describes speech as serving the purpose to ‘establish, to prolong, or to discontinue communication’ (Jakobson, Reference Jakobson and Sebeok1960: 355), i.e. it is language for the sake of contact. But if the point of posting online is most simply to make contact, then any post on social media, memes included, have an inherent phatic function.

When considering the examples 7 and 8 below which use the POV text-meme, we can see why a ‘merely phatic’ conclusion might be tempting. The POV text-meme is meant to provide a ‘point of view’, inserting the reader into the speaker's narrative about an event. The speaker rhetorically asks of the reader to consider that this propositional information is their own point of view; the text-meme as such originated as one that necessitated a video or an image that served as the POV. Examples 7 and 8 are however two extremely cryptic tokens of the POV text-meme. The hasty formulation in 7 suggests the speaker might have been drunkenly stating they are drunk. Example 8 is even more cryptic: what exactly is the point of view, with a co-text of just peepee poopoo? We can only speculate as to its intended meaning. Yet, even examples 7 and 8 go beyond the phatic function.

These examples still exhibit the indexical function, despite being not nearly as overt as other text-meme tokens in their indexicality, nor in what Jakobson called the emotive function of communication, which is language for the sake of speaker-centered expression. These text-meme tokens hardly reveal anything about the speaker or the context, and yet, the speakers index themselves as meme-users. This in itself reveals a particular membership belonging of the speaker. Zenner and Geeraerts (Reference Zenner, Geeraerts, Winter–Froemel and Thaler2018: 173) similarly describe image macro usage as marking an ‘ingroup code of the digitally literate’ (see also Bülow, Reference Bülow, Bülow, Koch, Krieg–Holz and Trost2023: 261).

Producing and sharing a meme is a highly performative act, even if practically a reflex. On the flipside of the indexical function is the interpersonal function, related to what Jakobson referred to as the conative function of communication, which is language for the sake of appeal. Memes constantly rely on an interface between addressers and addressees. They are self-referential and ultimately only meaningful by way of reception. That is, if reception is possible: the addressee has to be just as much in the know as the addresser. And if so, memes are then identified as intertextual references over word-for-word utterances, i.e., they are meant to be read as holistic chunks. For a meme to be considered a meme, prior and current usages are intertextually linked. Memes network with other internet material, sources of pop culture, and most importantly, with other memes (Laineste & Voolaid, Reference Laineste and Voolaid2017; Zenner & Geeraerts, Reference Zenner, Geeraerts, Winter–Froemel and Thaler2018; Wiggins, Reference Wiggins2019). This intertextual function is not at all specific to memes but rather to genres in general (De Beaugrande & Dressler, Reference De Beaugrande and Dressler1981). Example 9 from Twitter demonstrates the intertextual function by exaggeration:

This example of the me, an empath VERB-ing X text-meme is composed entirely of other memes as the co-text: the X urge to Y; my toxic trait is X; you/’re X, I/’m Y, we are not the same; and X: me and who? The speaker jumbles together these popular text-memes to make one seemingly nonsensical chain of text-memes (perhaps criticizing the overuse of the text-meme types). If these intertextual references to the text-memes were unknown, then the indexical and interpersonal point of example 9 would be missed. The meaning of example 9 also lays in the poetic function. Using Jakobson's terminology, the poetic function describes how the form of the message reflects ‘on itself.’ That is, when the message design makes the meaning, and when the form of the message is the point - apart from the meaning attached to the forms. Speakers do not only refer to the entities they talk about, but inevitably to the text-memes themselves (they are self-referential).

The notion of ‘message makes the meaning’ could also be explained in terms of the social function. This last function that we attribute to all meme types might be able to answer to a few special qualities of text-memes, such as their prolificity and tendency to change. Once memes are mimicked enough, they compete with other, newer, possibly even better memes. At that point, for a meme to survive rather than die, there is the potential to remix into something alike, but different (cf. ‘memetic drive’ in Blackmore, Reference Blackmore2000; or ‘intermemetic referentiality’ in Esteves & Meikle, Reference Esteves, Meikle and Atton2015; Wiggins, Reference Wiggins2019). In the case of text-memes, it is common for either the form of the meme or the conventions of the co-text to easily and quickly change. For instance, a well-established text-meme, the feminine urge to X, has been significantly revamped over time. On its so-called ‘adaptability’, the text-meme even got a feature in the New York Times in an article titled ‘The Timesian Urge to Explain a Meme (Kambhampaty, Reference Kambhampaty2021). Know Your Meme, a vast and up-to-date encyclopedic web source on meme tracking, estimates the first instance of what became a text-meme to have been the following Tumblr post (Know Your Meme, 2007-):

In early instances of the text-meme, the co-text would usually pair a transitive verb with no subsequent object. Semantically, the transitive verb would have little or nothing to do with what one associates with femininity, or ‘feminine urges’ nonetheless (e.g. stabbing). When in full throttle, there likely comes a split in imaginations of a fitting co-text of a text-meme type, and more variation in form too. Some interpret and use the text-meme ironically (example 11), and some more so at its face value (example 12). The feminine urge to X eventually allowed for an open slot to all kinds of urges, the modified construction being the X urge to Y (examples 13 and 14).

Inevitably, some meme-users would deem even the revamped constructions ‘dead memes’, despite whether some still use it (‘Dead meme: A meme that has become irrelevant or unfunny, often due to age or overly cringeworthy use’ [Urban Dictionary, 2017). The following speakers took to Twitter their meta-level complaints about the text-meme:

Divergent opinions on the repute of a meme seems unavoidable. Memes are like codes, in that they must balance being encoded enough (i.e. they have to be obscure enough, for only the in-group to get them), yet not too decoded (i.e. they should not be too obscure, in order for the in-group to get them). Text-memes necessitating obscurity may seem counterintuitive, if the argument is that text-memes are communicative utterances above all else. Why should text be obscure if functioning predominantly to communicate? A certain level of obscurity is critical for memes to ‘stick’. This is also a criterion that differentiates text-memes from snowclones or phrasal templates. For as much as memes function to be widely used and ‘mimicked,’ they also function to die, so that newer and cooler memes can be born. Relevance theory (Sperber & Wilson, Reference Sperber and Wilson1996) could give some kind of explanation for text-memes’ quick, successful and curious rise and fall as communicative forms among some internet users. In this approach, the notion of relevance is understood as being a trade-off between the positive cognitive effects to be gained from processing an utterance and the energy required for doing so. Memes are not superficially ‘relevant’, by means of that terminology, but are rather only relevant on a deeper level, like in their obscurity and humor. The process of working out meme's relevance is rewarding, in that the reader can rejoice once having worked it out, gotten the point of the intertextual references, and can be satisfied with being in the know. As soon as a meme is overused, there is no longer a reward in store. This is also a property typical of expressions that leave a gap between what is said and what is meant (cf. ‘euphemism treadmill’, Keller, Reference Keller1994; Pinker, Reference Pinker2002).

3.2 Narrative and therapeutic functions

An apt introduction to the distinct functionality of text-memes is to consider them stories, and speakers storytellers. It has been suggested that memes at large are analogous to storytelling and lore in how they retell a story repeatedly over time, spun by the speaker and their context (Esteves & Meikle, Reference Esteves, Meikle and Atton2015). This analogy remarks more on how memes as a genre work. What we suggest here is that text-memes in particular often a particular narrative function. Narrative, most simply put, is the record of someone telling someone else that something that happened, (for a pragmatic purpose). And, as is typical of narratives, there is always some kind of complication (Labov & Waletzky, Reference Labov, Waletzky and Helm1967).

Each instance of a text-meme entails two indexical jobs related to narrative concerns: speakers index themselves in relation to some piece of propositional content (expressed in their co-text), and speakers index themselves as meme-users, to express the former. In this way, text-memes can serve as self-mythologies, in the sense that the speaker – the narrator – recounts on an event, real or not, pinning themselves as the main character in and against the world. These self-mythologies tend to represent the complication component around which the structure of narratives revolves. Maybe for this reason, an initial impression of text-memes is that they come across self-centered. This is a legitimate impression, because text-memes, even when the subject is not the speaker, are often collections of personal anecdotes and morals related to the experiences of speakers (remember examples 1 and 2, where speakers attempt to set the record straight that they are ‘empaths’). And very often, the topics of such text-memes are overshares or taboos:

Examples 18 and 19 from Twitter demonstrate what is meant by text-memes as mini-narratives: they set up a scene and describe an event representing the complication. In 18, the literal narrative is how the speaker reflects on the death of their pet cats and sarcastically attests they are currently ‘very stable’ despite also contemplating death; in 19 the speaker narrates humming while being broken up with. Beneath the hypothetic and hyperbolic tone common of POV text-memes, 18 and 19 likely reveal actual sentiments felt by the speakers: feeling devasted, or, being indifferent to something usually considered devastating. We mentioned that all meme usage indexes speakers as meme-users, but the narrative quality of text-memes takes self-indexation one step further. The therapeutic function might explain the question as to why this is. Why tell the story, and why use a meme to tell it?

We suggest that text-memes very often exhibit a therapeutic function, in which speakers use them to cope or engage with a wide-range of emotions:

The POV affords speakers a particularly easy, indirect outlet to speak from the ‘you’ rather than the ‘I’ perspective. It is a fixed formula to out oneself, own up to or open up about, for instance, feeling alone or basking in childhood nostalgia.

Global, collectively-felt events are also a favorite topic among text-memes, which play into the therapeutic function. A study on 9/11 and Iraq war pictorial memes or image macros, such as the Mission Accomplished motif (Silvestri, Reference Silvestri2018), brought to light how memes can contribute to past or current social and political events by synthesizing them and raising awareness for them, and by doing so, take on immense responsibility for how online communities come to view the events (cf. Crovitz & Moran, Reference Crovitz and Moran2020). How on one hand, it is hopeful that memes are segues to activism, and on the other, there is concern that memes offer a lightheartedness and humor which can make it a challenge to discern serious matters from the silly or simplified messages typical of image macros. It would be interesting to see how text-memes compare in this sense; for instance on the topic of the SCOTUS ruling to overturn Roe V. Wade:

Inherent to the more expansive, textual form of text-memes and the difference in their communicative abilities, we could at least initially suggest that compared to political image macros, text-memes could by nature allow for easier, more thoughtful argumentation and activism. Example 22 demonstrates this ability of text-memes well. Example 23, quite crudely, reminds us that text-memes are still memes, and that the extra creative license of text-memes is somewhat of a double-edged sword, just like with image macros. Meme-users can speak out against injustice just as much they can make light of it. Humor is of course an important feature in memes as entertainment devices, also as coping mechanisms for the speakers themselves, as might be the case in the examples above. To Silvestri's (Reference Silvestri2018) point, meme users indeed become stakeholders in important geo-political affairs. While this notion seems promising on some fronts, for instance in countering political apathy among young people, the boldness and comedic delivery which has potential to be insensitive is an issue of concern when it comes to serious, unfunny issues. In their efforts to act or cope with current events, it is the individual meme-user's responsibility to decide how they set the tone.