It’s difficult to imagine the internet today – and social media platforms especially – without internet memes. These ubiquitous image-text combinations pair recognizable but modifiable forms with consistent meanings and contexts of use, allowing communicators to bring across viewpoints and elicit responses quickly. Memes can be very easily made and adapted thanks to meme-generating platforms, so using internet memes to provide a comment on some situation need not be unduly complicated – in fact, we often see people online responding with amazing speed to some news item. We should add that summarizing in words what memes communicate across the interlocking attitudes they incorporate is much more complicated, and much less appealing, than using the conventionalizing, ‘construction’-like combinations of text and image memes typically deploy (Dancygier & Vandelanotte Reference Dancygier and Vandelanotte2017b). Indeed, internet memes seem tailor-made to express opinions which they do not literally or straightforwardly state, but which instead emerge from the viewpoint structure the meme builds up. Building on humorous forms of indirectness or incongruity, they parcel up often emotional messages in more attractive, playful, and therefore more impactful ways than just words.
From marginal beginnings, internet memes thus have become part of the fabric of everyday communication online, even becoming a kind of ‘second language’ communicators can have different degrees of proficiency in. Scholars of communication and rhetoric have proposed ways of dealing with what memes are, how they emerge, and how they contribute to public discourse (e.g. Jenkins Reference Jenkins2014, Shifman Reference Shifman2014, Wiggins & Bowers Reference Wiggins and Bret Bowers2015, Milner Reference Milner2016, Wiggins Reference Wiggins2019). In the process, they have clarified differences with the original conception of memes as coined by Dawkins (Reference Dawkins1976) – where he had memes in mind as a unit of cultural replication, as a counterpart to genes, with examples including, for example, catchphrases, tunes, fashions, architectural styles, and so on (see also Blackmore Reference Blackmore1999). The emerging scholarship on internet memes has stressed the active human agency and creativity in deliberately devising modifications and remixes to existing ‘spreadable’ cultural artefacts. Any idea of passive transmission combined with random mutations from the ‘gene’ source frame is clearly not an appropriate model with which to understand the frenetic energy with which meme users exploit internet memes’ meaning potential by adding their own textual and visual tweaks.
Much scholarly work has focused on the social and political meanings and ramifications of meme usage (e.g. Huntington Reference Huntington2016, Milner Reference Milner2016, Ross & Rivers Reference Ross, Rivers, Schill and Hendricks2017, Denisova Reference Denisova2019, Paz et al. Reference Paz, Mayagoitia-Soria and González-Aguilar2021), and interdisciplinary work is trying to address the vexed question of what determines a meme’s success and longevity (e.g. Johann & Bülow Reference Johann and Bülow2019), where sometimes brief memetic moments can nevertheless have a ‘long tail’ (Smith & Copland Reference Smith and Copland2021). In discourse studies of various persuasions, pragmatic aspects of internet memes, touching on questions such as common ground, intertextuality, and humorous incongruity, take centre stage (e.g. Yus Reference Yus2018, Xie Reference Xie2022, Attardo Reference Attardo2023).
While book-length studies like those of Shifman (Reference Shifman2014), Milner (Reference Milner2016), and Wiggins (Reference Wiggins2019) investigate and elucidate a broad range of questions, we felt that a fuller analysis from a linguistic perspective was missing. In line with our earlier work (Dancygier & Vandelanotte Reference Dancygier and Vandelanotte2017b), we take an approach inspired by construction grammar and cognitive linguistics, which we share with analyses proposed by, among others, Lou (Reference Lou2017), Zenner & Geeraerts (Reference Zenner, Geeraerts, Winter-Froemel and Thaler2018), and Bülow et al. (Reference Bülow, Merten and Johann2018).
The variety of internet memes and meme types is considerable, and we will not attempt an exhaustive classification or typology of internet memes. We are more interested in the meaning-making mechanisms involved, which rely on, expand on, and add to existing linguistic mechanisms and patterns, in major image-text memes such as so-called Image Macro memes, labelling memes, or grid memes. In general, our goal is to describe how memetic forms create a rich and cohesive system serving the needs of meaning construction.
We also want to comment on the issue of the inherent datedness of memes which are older than a week. One might argue that memetic communication, in terms of both meaning and form, is so fleeting and time-sensitive that any interpretive or theorizing efforts, such as ours in this book, are bound to produce something outdated. Indeed, we are aware that some meme formulae go out of common use (e.g. it might be hard to find very recent uses of the Said No One Ever meme). But the emergence and subsequent disappearance or evolution of meme forms is a process which is inherently interesting and important. We might refer here to the work done by Lou (Reference Lou2017) which traces the evolution of the so-called when-meme, from more wordy forms (e.g. That Awkward Moment When) to the simplest, current one, which remains popular and useful. Memes are still coming into their own as a communicative genre, and they evolve at lightning speed, but the process is important and instructive. It was observing this ‘living lab’ of multimodal ‘grammaticalization’ that convinced us that memes are a topic worthy of a book-length analysis, not only because of what they do day-to-day – such as, providing the until-recently unavailable tools for immediate and interactive exchange of opinions – but, perhaps primarily, because of the vast possibilities that memes open by using minimal form to evoke and express complex opinions. The ‘minimalist’ side of memes allows them to connect internet users without any spatial or personal barriers. Understanding the mechanisms that make this possible is one of the goals of this book.
Because of the vast (and growing) variety of forms in use, memes are very difficult to define as a coherent communicative genre. A commonly used definition of internet memes is that proposed by Shifman, who understands an internet meme as ‘a group of digital items sharing common characteristics of form, and/or stance, which were created with awareness of each other, and were circulated, imitated, and/or transformed via the Internet by many users’ (Shifman Reference Shifman2014: 41). This definition rightly stresses the active, creative participation of communicators. It also makes the point that internet memes do not exist in isolation but form part of identifiable, structured groups of items; not every one-off joke or comment that happens to be shared on the internet is automatically a meme (for discussion, see Vandelanotte Reference Vandelanotte2021: 160–4). To categorize something as an internet meme requires some formulation of what ‘group of digital items sharing common characteristics’ it belongs to. Throughout this book, we will be considering a range of such characteristics, with a goal of explaining the major formal choices Meme Makers face in terms of primary meaning-making strategies.
Before delving into a description in further chapters of how various memetic formulae use language and images to express opinions and portray experiences, we would like to comment more broadly on several aspects of memetic discourse. We start (in Section 1.1) by clarifying our approach to the concept of ‘multimodality’ and presenting memes as a specific genre of multimodal (image-plus-text) artefacts. Next, in Section 1.2, we preview the formal and interpretive parameters which provide a broad frame to the analyses throughout the book. Finally, in Section 1.3, we outline the subsequent chapters in broad strokes.
1.1 Memes as Multimodal Artefacts
The overwhelming majority of memes we discuss in this book rely on a combination of text and image – thus they fall naturally into the category of multimodal artefacts. However, we argue that memes represent ‘multimodality’ in a rather unique way.
There are two main ways in which ‘multimodality’ has been studied: the semiotic approach considers artefacts using various semiotic modes (such as sound, image, colour, light), often combining them with text, while the embodied interaction approach studies communicative channels such as gesture and eye-gaze and the ways they co-align with spoken discourse. In past work, we have suggested that these different traditions can be drawn on without contradiction, recognizing that multimodality is ‘a varied, but nevertheless cohesive phenomenon’ (Dancygier & Vandelanotte Reference Dancygier and Vandelanotte2017a: 372).
The embodied interaction approach studies multimodality ‘in interaction’, where the context is spoken or signed discourse, and embodied behaviour supports the linguistic forms chosen. The correlation between embodied aspects of communication is the primary interest in such studies (for an overview, see Feyaerts et al. Reference Feyaerts, Brône and Oben2017), and so we are finally beginning to understand the phenomenon of face-to-face communication as it typically happens. And we are learning even more from seeing how the co-alignment of linguistic and embodied resources speakers bring into communication creates complex, and inherently, or even inescapably, multifaceted behaviour. The main point of the approach is not just to parcel out meanings by assigning them to different interactive modalities, but to also demonstrate the necessary integration of speech, gesture, eye-gaze, facial expression, and so on. Our treatment of memes follows in the footsteps of this line of research, by stressing the ways in which memetic text-and-image participate in an integrative process, rather than parcelling meaning out to different expressive modes.
The proper understanding of the importance of the interaction paradigm is further complicated by mimetic aspects of communicative behaviour that Clark (Reference Clark2016) has described as ‘depictive’ – including depictive gestures, but also tone of voice, imitation of music, enactive use of the body, and many other contexts. What Clark’s work makes us realize is that the interactive resources we build on in spontaneous communication are used for much more than the current communicative context. The gesture or the facial expression used may be aligned with what the speaker him/herself wants to communicate to the hearer, but it can alternately be aligned with a represented subjectivity (speaker, singer, writer, Meme Maker, painter, fictional character in a story, and so on). What such situations suggest is that none of the embodied modalities themselves (gesture, tone of voice, or body posture) tell us all we need to know about the act of communication developing before us. We would like to return here to the example we discussed in earlier work (Reference Dancygier and Vandelanotte2017a), where a stand-up comedian uses his body, face, voice, and gestures to tell a story in which he represents himself (the former-participant-plus-current-narrator), but also a character, and an imagined stand-in for that character (whose behaviour he needs to act out in a depictive way to make it clear why he finds the character’s behaviour offensive). Depiction is the core of this comedy act, all the communicative resources are used, but we would be rather hard-pressed to explain how the specific communicative modes are being used, even though listing them would be easy – because none of them exists separately from the multiple levels of the story being told. Re-enacting another participant in a story being told is often necessary in spoken storytelling, but such a communicative act cannot be satisfactorily described solely by reference to the independent modalities. If there were a term in multimodality studies such as ‘spoken performative humorous multi-voice narrative mode’, that would be a satisfactory term perhaps, but it would have the huge disadvantage of being just one of many such options. In the context of memes, which can also, to a degree, rely on representing depictive modalities, we attempt to trace specific patterns which combine visual and textual elements, but not as random combinations. On the contrary, we track regularities of form which target specific patterns of meaning emergence.
As we show, depictive uses of embodied modalities play an important role, so that (often exaggerated) facial expressions, gestures, or body postures are common in memetic images as signals of emotional or experiential content. But complex artefacts such as a stand-up comedy act are hard to compare to the formal simplicity and mostly static nature of memes. Memes build on aspects of communication that ‘multimodality in interaction’ research studies in detail, and they can represent bodies and facial expressions as tokens of experiences, but they can also evoke experiences and emotions without representing experiencers as such, and choosing other visual ‘stand-ins’ instead (as we show in Chapter 2). Our project thus builds on the ‘multimodality in interaction’ approach, but our proposed analysis is not strictly dependent on the affordances of the human body.
The semiotic approach to multimodality, for comparison, focuses mainly on artefacts representing more than one expressive mode (e.g. image, sound, colour, etc.), rather than spontaneous interaction (see, e.g., Kress & van Leeuwen Reference Kress and van Leeuwen2001, Bateman Reference Bateman2008, Kress Reference Kress2010, Bateman Reference Bateman2014, Jewitt Reference Jewitt2014, Nørgaard Reference Nørgaard2019, Hart Reference Hart2025). Our proposed theoretical stance uses the insights from such approaches, but we are looking at memes as artefacts which primarily rely on images and text, though we will make occasional reference to the more dynamic, video-based genre of TikTok memes, as and when appropriate. While we are theorizing the ways in which image and text create meaning in memes, we are going beyond seeing image and text as independent modes of expression.
1.2 Memes as Multimodal Constructions
We argue throughout the book that memetic use of images and words does not use independent expressions from independent communicative modes to put them in a semiotic or figurative correlation. More importantly, perhaps, we are cautious about treating words as a ‘communicative mode’ on a par with image or sound. Rather, we want to treat them as tokens of language use structured by the formal features which languages rely on – we usually call them ‘grammars’. Language is not just a ‘mode’ – it is a complex system of concepts, lexical as well as structural, which can also structure newer uses in a variety of internet discourses and digital media. What memes do, we argue, is use linguistic forms in combination with images, while fully relying on the linguistic systems of form-meaning correlations. This reliance does not have to yield typical language use – on the contrary, it creates new uses, which may alter some standard formal rules, but which are not random or just clipped to fit in the memetic image.
For example, in a series of Image Macro memes known jointly as Good Girl Gina memes, one sees an image of a very pleasant-looking young woman (we will return to the example in Chapter 3, Figure 3.2). The woman’s memetic name is Gina, and she functions as an epitome of a perfect girlfriend. In the top part of the meme there is a line of text, such as Gets mad at you; another line of text occupies the bottom of the image – tells you why. The lines of text do not specifically complement or figuratively reconstrue the image. The relation between image and text in such memes is not about two modes representing the same thing or using an image as an illustration of the text. The image is not uniquely associated with the text included, as it is the standard image for all Good Girl Gina memes – you recognize those memes because of the image. Also, the true identity of the girl is not under consideration – she just stands for the idea of a perfect girlfriend, very much in the way in which the expression a perfect girlfriend refers to a category of female human beings who look and behave in generally pleasant ways.
Now, the text in this meme (Gets mad at you … tells you why) does several linguistically interesting things. First of all, it suppresses the need for an expression of the subject argument – contrary to the ordinary grammar of English – and does so, we argue, because the identity of the subject is sufficiently represented by the image. The image is not seen as a photo of a specific girl, but as a compositional part of any Good Girl Gina meme. Also, the text, while consisting of two clauses, cannot be read as a fully expressed construction. It relies on generic predictive constructions (If she gets mad at you, she tells you why; When she gets mad at you, she tells you why; She gets mad at you and tells you why; etc.), and so it is evocative of a language structure expressing a specific causative/predictive meaning (Dancygier Reference Dancygier1998, Dancygier & Sweetser Reference Dancygier and Sweetser2005). Even if the construction is structurally simplified – not requiring the subject noun phrase (NP) or a conjunction – the linguistic pattern is still fully upheld. The clauses cannot be used in a different sequence, and the top clause cannot be completed with another construction (such as told everyone why). The image and text participate in a tightly correlated pattern, a construction, even if the construction is not solely reliant on language – it prompts a new, multimodal construction (complete with pre-filled parts, such as the image), grammatical restrictions (subject suppression, allocation of the adverbial clause to the top of the meme and the main clause to the bottom of the meme). In terms of its meaning, the meme specifies a behaviour any perfect girlfriend should consider natural, while also intersubjectively expressing some mocking criticism of some women, who presumably refuse to discuss the cause of their anger when they get upset, and/or perhaps prompting awareness in Meme Viewers of their own complicity in sexist attitudes. We are not looking at a ‘combination of two modes, image and text’, each contributing separately and only being combined in a final interpretive step. Rather, we are looking at a multimodal construction, a form-meaning pair, with specified categorial meaning and intersubjective grounding.
We argue for such an approach to memes throughout the book. We will discuss a number of types of memes to propose generalizable observations about the specific formal and meaning parameters involved. We will consider a range of formal questions: the constructional use of the entire space of the meme, the specific constructional role of images, the constructional emergence of expressive stances, the roles of token snippets of direct discourse in such constructions, the use of pronouns, and, generally, the linguistic adjustments memes rely on in the emergence of multimodal memetic constructions.
In this book we will develop ideas anticipated in our earlier publications, where we already argued that memes can be seen as multimodal constructions (Dancygier & Vandelanotte Reference Dancygier and Vandelanotte2017b) – in parallel to linguistic constructions, as described in Construction Grammar. We will continue to rely on the simplest definition of a construction as a form-meaning pair. In construction grammars, linguistic form can be a type of syntactic structure, such as a Ditransitive Construction (e.g. Bob gave Alice a book) or a comparative expression The X-er the Y-er (e.g. The sooner the better). Such a broad approach to constructions has yielded several more specific frameworks (Hoffmann Reference Hoffmann2017b), offering descriptions of form and meaning in one coherent analytical approach. We now propose to extend the approach to an analysis of internet memes, showing how they express stable meanings on the basis of formal configurations containing text and image. As we will show throughout the book, memetic constructions range from very open-ended patterns to highly restricted formal ones.
1.3 The Structure of This Book
In the next chapter, Chapter 2, we first consider some correlations between memetic constructions and select figurative meanings, showing how our approach differs from existing multimodal metaphor approaches. As a case in point, the chapter presents an analysis of when-memes as relying on similative patterns of meaning, and also extends this discussion to include the family of If 2020 Was X memes.
Chapters 3 to 5 then focus on major structuring strategies across a large number of memes: the role of the image (as an entrenched ‘macro’, or more ad hoc and ‘non-entrenched’) and its combinations with text (Chapter 3); the use of labels on top of parts of an image (Chapter 4); and the use of various kinds of grids as a particularly popular presentational template (Chapter 5). These are not necessarily distinct ‘types’ of memes, since these strategies happily combine, so that for instance an Image Macro may itself be a four-way grid and be labelled up, as we’ll see. Based on the many examples we’ve looked at, we do think these three types of structure capture a lot of the fundamental formal variety of memes.
Chapter 3 introduces the distinction between entrenched images or Image Macros (IMs) and Non-Entrenched Images (NEIs), and focuses most of its discussion on examples involving IMs that feature the characteristic Top Text (TT) and Bottom Text (BT), such as the One Does Not Simply and Good Girl Gina memes (ODNS and GGG). It shows how these IM memes allow Meme Makers to categorize experiences very quickly, efficiently, and (if successful) humorously, adding further examples to such categories as ‘futile undertakings that are impossible to achieve’ (ODNS) or ‘virtuous behaviour of highly considerate women’ (GGG), thanks in large part to the frames evoked visually. It also discusses aspects of the construction grammar approach to language, as applicable to these meme constructions, including specific constructional properties of GGG memes and the constructional networks they fit into.
Chapter 4 turns to labelling memes, where again some images may develop into full-blown IMs, while others remain non-entrenched. Here, the textual component is different from both when-memes (discussed in Chapter 2) and from the typical IM memes with TT and BT. Instead, in typical cases of labelling, parts of a visually depicted scene are labelled with words or phrases which do not describe anything in the image, but instead collectively call up an entirely different frame. Well-known examples the chapter discusses include the Is This a Pigeon? meme, and the celebrated Distracted Boyfriend meme (DBM), showing a man turning over to admire an attractive passing woman (dressed in red), while the woman (in blue) whose hand he’s holding looks on indignantly. This scene of a change in attention and preference – a choice for a new and attractive opportunity – gets to be applied to unrelated choices and new preferences. Labelling itself, as we show, can sometimes be visual again. Overall, we stress the constructional properties of DBM – with strong argument structure-like properties – alongside the role of embodied features (emotions and attentions expressed in facial expressions and posture) and the figurative, similative meaning often arrived at compositionally.
Chapter 5 discusses a variety of presentation formats involving grids, which we tend to scan left to right, top to bottom. Some grids are scalar, structuring a graded sequence of experiences (for instance, formality of language, in Tuxedo Winnie the Pooh memes), or even correlating two scales as in the Political Compass meme we discuss. Others involve contrasts (as in Drake vertical grids), or structure discourse exchanges and narrative sequences in grids (among our examples here is the Anakin and Padmé meme). These different uses of grids allow Meme Makers to present and confront different behaviours, stances, and attitudes which Meme Viewers take as prompts from which to construe a coherent, typically ironic, viewpoint.
The next series of chapters turns to issues which relate less to the overall presentational formats of memes, but to sub-forms like pronouns (Chapter 6) and (apparent) quotation and dialogue (Chapters 7 and 8) appearing in them, and to formal play and experimentation itself (Chapter 9).
Chapter 6 considers the use of pronouns, and how they relate to such roles as Meme Maker, Meme Character (depicted in a meme’s image), and Meme Viewer (i.e. the intended ‘reader’ of a given meme). One illustration of how odd pronouns actually behave in memes is to consider the use of I, which does not refer to the Meme Maker, but is used to represent embedded discourses attributed to a depicted Meme Character. Just as curious is the use of me, in patterns such as Me Verb-ing, or Me/Also Me, which apparently instruct us to look for Meme Maker in the meme’s image, which in fact shows an unrelated Meme Character (possibly non-human, like an animal), such that the depicted character represents the experience of the Meme Maker. Such examples show that deixis is used in unusual ways in memetic discourse, to support the expression of viewpoint and stance targeted in the meme, rather than to identify specific referents.
Chapters 7 and 8 home in on uses of memetic quotation – cases where (apparent) Direct Discourse, and even (fictive) dialogic exchanges, are featured in memes. Chapter 7 first surveys such usage in cases that are closer to recognizable existing linguistic constructions involving verbs such as say, tell, and be like, but adding further constructional specifications in their memetic applications, thereby yielding very specific meanings. Forms analysed include Said No One Ever, It’ll Be Fun They Said, And Then He Said, What If I Told You, and Be Like; the latter in particular sometimes combines with very complex content being ‘quoted’ or demonstrated, as we illustrate. Chapter 8 expands the discussion to quotations and what we call ‘dialogue labelling’ in other memes, not featuring explicit reporting verbs, but relying on depiction of interlocutors, interpretation of embodied behaviour, and sometimes quotation marks to signal the embedded Discourse Spaces, and viewpoints exchanged, in them. We also analyse a range of discourse patterns building on the basic Me/Also Me introduced in Chapter 6, and round off with the Repeat after Me meme.
Having surveyed many elements that determine the typical forms and meanings of memes, Chapter 9 turns to memetic experimentation. Meme blends, metamemes, or cases of ‘memeception’ (or recursivity in memes) all manipulate aspects of form to create new meaning effects. Antimemes, on the other hand, do not alter the form, but change the viewpoint structure and so, the meaning. Some memes, finally, appear to enjoy memetic form for form’s sake, and border on art forms; the so-called Loss meme is our main example here.
Chapters 10 and 11 both look ‘outwards’, to broader contexts of use where memes and meme-like forms circulate: social media platforms (Chapter 10) and advertising (Chapter 11). If, in earlier chapters, we mainly studied memes in a ‘platform-neutral’ kind of way, in Chapter 10 we consider more platform-specific forms exploiting possibilities such as the ready integration of emoji on X/Twitter or the integration with audio and video on TikTok. We also pay attention to the co-construction of memetic discourse by multiple discourse participants in online exchanges. Overall, we suggest that the easy transfer across platforms and modes reveals a kind of memetic mindset in which discourse takes shape online, even where this does not necessarily involve fully formed or identifiable memes in a strict sense.
A similar observation pertains to the advertisements we study in Chapter 11, where our most interesting examples don’t so much directly borrow a fully formed, recognizable meme to reuse it in an ad (though this, too, is sometimes done). Instead, really successful memetically inspired ads partly borrow from existing meme codes, such as when-memes or the Sections of meme, and adapt these creatively to suit the persuasive goals identified. This again suggests to us that aspects of the grammar of memes are affecting other forms of communication.
We tie together and reflect on this emerging grammar of memes in the closing Chapter 12, where we also revisit some of the questions first asked in this opening chapter, about why linguists should study memes, or how the specific kind of multimodality in the memes we studied differs from other multimodal genres.
Colour versions of all the figures featured in this book are available under the ‘Resources’ section of the website [www.cambridge.org/languageofmemes]. We will occasionally draw attention to aspects of colour and composition that may be hard to discern in the black and white print version, but we hope the availability of this online counterpart will make such points easier to follow.
