Fieldnotes: English and French in Japan

Blog | Research Centre for Multilingual Practices and Language Learning in Society

fresh winds

It’s late November 2019 and I just returned from a two-week trip to Japan. I admit that on journeys such as these, the fieldworker in me tends to take over at times from the tourist or visitor, for Japan remains a stunningly interesting place for issues of language and globalization. I discussed a number of Japanese examples already in The Sociolinguistics of Globalization, now almost a decade ago, and my recent visit added new and rich linguistic landscape data to the voluminous collection I have built over the years.

Let me quickly reiterate the argument I developed earlier regarding the phenomenology of globalized languages such as English and French. I started from the observation that linguistically ‘incorrect’ forms of English and French account for a lot – perhaps most – of what we see of these languages in the world. It is too easy to dismiss such forms of language…

View original post 780 more words

Trump’s viral errors


Jan Blommaert

Introduction: a perennial agenda

The discipline we now call sociolinguistics has throughout the 20th century systematically maintained and elaborated two connected issues.1 Note that ‘sociolinguistics’ as it is now called is an innovation of the 1960s, when scholars (mainly in the US) started using the label to distinguish themselves and their work from that of the Chomskyan paradigm in linguistics, and to emphasize continuity with an older paradigm incorporated in anthropology and exemplified in the tradition started by Franz Boas (Darnell 1998; Hymes 1992; Bauman & Briggs 2003). It is in this longer tradition that the two connected issues were given a definitive shape. The issues are:

  • the principled equality of all languages and
  • their factual inequality.

Taken together, these issues define sociolinguistics as a discipline concerned with diversity, but in a particular way.

The first issue, unpacked, has to do with the observation that every language, when seen in its concrete social context, is ‘perfect’: its resources enable members of the community of usage to express all possible meanings and fulfill every social-communicative function. In Benjamin Lee Whorf’s (1956) famous view, every language incorporates, expresses and shapes the worldview of those who use it, and those so in its very structure (cf. also Silverstein 1979). The issue was clearly articulated in Boas’ seminal Introduction to the Handbook of American Indian Languages (1911, also Boas 1928) as well as in Sapir’s groundbreaking Language (1921). It became the epistemological, moral and political point of departure as well as the battle cry of generations of sociolinguists, and it defined the linguistic scope of the new discipline.

The second issue defined the battlefield of sociolinguistics. Given the in-principle equality of all languages, how come so many languages are factually considered inferior to others? Why are speakers of so many languages oppressed and marginalized, why do we make distinctions between ‘standard’ and ‘substandard’ varieties, why do we consider dialects features of backwardness and remnants of a pre-modern past? Why do we attach stigma to some accents in a language and prestige to others – when both are linguistically equivalent? And why are such distinctions codified in language policies and cast in even more robustly policed language ideologies enabling and sanctioning discriminations in which linguistic differences are turned into sociolinguistic inequalities?

This second issue, certainly from the 1960s onwards, defined the social scope of sociolinguistics, and it can be summarized in one word: stratification. And there were precursors: ‘salvage linguistics’ – the study of languages threatened with extinction – emerged out of an awareness that such languages would disappear not because of their intrinsic inferiority compared to, say, English or Spanish, but because of the fact that increasing marginalization of the users of such languages would ultimately eliminate the languages. And such forms of marginalization often included a strong stigma – a perceived, ideological inferiority – for the languages and language varieties as well. They were not qualified as ‘languages’ but as ‘dialects’, ‘speech’, ‘jargons’, ‘sabirs’ or simply ‘barbarian’ and ‘primitive’ (cf. Fabian 1986a, 1986b). Certainly when these language were not accompanied by an identifiable writing system, they were considered to be expressions of the innate and therefore general inferiority of their users.

As soon as a branch of scholarship emerged carrying the label of sociolinguistics, both issues merged into an agenda, expressed and developed in the work of the leading scholars of the first generation of sociolinguists. Forms of sociolinguistic diversity, ranging from AAVE in the US (Labov 1970), native-American stories (Hymes 1983), ‘nonnative’ Englishes in the US and the UK (Gumperz 1982) or working class accents in the UK (Bernstein 1971) and minority-majority multilingualism (Fishman 1971) were shown to be the object of intense discrimination, notably in education (the focus of e.g. Labov 1970; Bernstein 1971; Hymes 1980). Such forms of discrimination had social, not linguistic causes, and their analysis as linguistic phenomena needed to be set in a context that was at once structurally formed as well as synchronically enacted, often with predictable outcomes due to the pervasive and enduring influence of policies and language ideologies rationalizing (and rendering ‘natural’) the stratification of sociolinguistic regimes (cf. Kroskrity 2000; also Bourdieu 1991). Increasing diversity, for instance due to globalization processes, appeared to merely increase and complicate sociolinguistic inequalities (cf. Blommaert 2005, 2008, 2010; Arnaut et al 2016).

This very quick run through a century of sociolinguistic history takes me to the point of departure for this contribution. While we must take stratification as the basic engine behind the dynamics of sociolinguistic systems, the actual forms of stratification have become somewhat less predictable due to what we call the online-offline nexus: the fact that large parts of the world’s population now organize and live their social lives online as well as offline, with both zones of social life, so to speak, being mutually influencing (cf. Blommaert 2018). Offline practices are profoundly influenced and altered by online infrastructures and vice versa, creating different sociolinguistic economies – patterns of resource distribution, general formats for conducting communicative actions and forming communities – and repertoires adjusted to such changed economies.

A simple example can suffice to illustrate the changes: emojis have become part of the everyday repertoires of visual design of many millions of language users across the world and (while not ‘belonging’ to any language in particular) have rapidly acquired specific, conventionalized communicative functions and effects. Philip Seargeant (2019) perceived this development as nothing short of an ‘emoji revolution’. Now, emojis are not part of most language learning curricula – their usage is often explicitly proscribed in language classes – and their usage is ‘chronotopic’, confined to particular and situated timespace arrangements such as scripted online interaction, advertisements and popular culture (Kroon & Swanenberg 2019; cf. also Blommaert 2015). But within such chronotopes, they are, if you wish, features of ‘standard’ language with a tremendous, transnational and translinguistic scope of usage and variant productivity (e.g., when the fully-formed smiley emoji is not available, it can be realized by means of other typographic signs such as ‘:-)’).

Similar things can be observed with respect to hashtags – the ‘#’ sign – as well as with the global spread of the ‘@’ sign to denote time and place as well as addressees in a wide range of scripted messages. Both are widely used in complex functions, and such usages display strong degrees of normativity (Blommaert 2020). Observe that such signs do not remain online but can be transported to offline chronotopes as well. Hashtags, notably, are widely used in demonstration banners, posters and flyers as well as on clothing. Hashtags have become a near-global sign indexing ‘message’ in general. At a higher-scale level of communicative economies, we see how online social genres such as tweets or Instagram updates have become incorporated into domains of power and prestige – they have become firmly integrated into political campaigns, for instance, and now compete for prominence with older genres such as the politician’s public rally speech or the newspaper editor’s op-ed article.

Restratification in the online-offline nexus

All of this means that the normative world in which sociolinguistic resources get their place and value allocated needs to be reconsidered. The expansion of the infrastructures for communication have inevitably gone hand in hand with an expansion of the ‘centering institutions’ described by Michael Silverstein (1998: 404; also 1996) as the real or imagined sources of normative authority for social-communicative conduct to which people orient while communicating, and through which their conduct is appraised and ratified (cf. also Agha 2007). The result is a complex polycentric sociolinguistic system, i.e. an unstable, dynamic and open one in which gaps and overlaps, conflicts, contradictions and nonlinear outcomes are the rule rather than the exception (cf. Blommaert 2016).

Of course, this statement, as soon as it is formulated, appears pedestrian, almost truistic. Perhaps sociolinguistic systems were always complex ones (as prefigured by e.g. Bakhtin and Voloshinov when they emphasized dialogism and heteroglossia), and perhaps the only virtue of the online-offline nexus is that it takes this simple given into the spotlight and makes it inevitable. But even so there is a moment to be captured, for this insight forces us towards another imagination of the major vectors and patterns of stratification and restratification – away from simple top-down models of imposed and carefully engineered hegemony (as in early studies on language policy and language planning, e.g. Eastman 1983), from stable binaries of majority and minority languages at societal level  with linear effects of linguicide looming (e.g. Phillipson 1992) and from studies of forms of language mixing as aberrations of a supposedly homogeneously monoglot norm (e.g. Myers-Scotton 1993). Theoretically as well as empirically, we need to see the normative valuation of sociolinguistic resources and of the modes of communication they shape, as well as the stratifying outcomes of such valuations, as sets of different effects spread over and caused by a range of actors and involving several very different types of activities, some of them involving high degrees of agency and others low degrees, some of them obviously revolving around human decision-making while others involve algorithmic technologies in crucial aspects of the process. Simply calling all of this ‘power’ may be comforting shorthand, but does not do justice to what actually goes on. The question is really: which specific forms of power generate stratifications and restratifications in online-offline situations.

I shall try to answer this question by means of an extended case analysis. I can offer a spoiler at this point. We shall see that the online language of the powerless can be appropriated by the powerful precisely because it is transgressive and evokes strong moral condemnation from powerful groups, and that such curious reversals of conventional sociolinguistic stratifications can algorithmically be turned into a partisan ‘majority’ norm in a fragmented public sphere. The case I have chosen involves the most powerful person on earth: the President of the United States of America. It involves English, the world’s most stratified language because it is the most globally distributed one. And it involves the sociolinguistic object most sensitive to normative judgment: orthography.

Trump on Twitter

There is a very strong cultural assumption in societies such as ours, in which the most powerful people are also the sociolinguistic elites: they are expected to command the most advanced and highly valued communicative resources. When they talk, they are fluent and eloquent in ‘standard’ varieties of the most prestigious languages; when they write they write elegant and elaborated texts in accordance to the strictest rules of grammar, genre and orthography. And in all of this we expect these people to be coherent, make sense and preferably sound intelligent. This assumption rests on robust sociological grounds, as the oeuvre of Pierre Bourdieu demonstrated: dominant groups in society are the guardians of norms in the field of culture as well as in the field of language, and when a variety of language is called ‘accentless’, we are actually facing the most prestigious accent – that of the elites (cf. Bourdieu 1987, 1991; Agha 2007). It is further undergirded by an army of professionals supporting the powerful in their communicative work – from speech writers to communication advisors and social media staff – and ensuring the best possible discursive products whenever one needs to talk or write.

There is no doubt that Donald Trump can draw on the services of an exceptionally large and exquisitely equipped army of such communication specialists. He could already do so before his election to the US presidency in 2016, and it is safe to assume that he could benefit from the services of the most outstanding members of the profession after he moved into the White House. Yet, since the very beginning of his electoral campaign, Trump’s discursive idiosyncrasies became the object of intense public discussion.

Of course, he had big shoes to fill as a communicator, being the successor to one of modern history’s most accomplished public orators, Barack Obama. But then, Trump was not the first US president to be targeted for public communication flaws. Obama succeeded George W. Bush, a president whose incoherence and inarticulateness in public speech had become the stuff of legends (see Silverstein 2003; Lempert & Silverstein 2012). Bush, with a Texas drawl, would fail to get the pronunciation of relatively simple words and names (such as ‘Europe’) right, he would produce incoherent ramblings in answers to reporters, would deliver contradictions in terms and so forth. Such communicative flaws were widely perceived to be deeply embarrassing for almost anyone associated with Bush, and as a sign of a character flaw called ‘questionable intelligence’ for Bush himself. But there still was the army of communication professionals, able to prevent the unfiltered and unedited presidential ramblings from becoming US policy, and able to turn incoherent statements into coherent (or coherently explained) ones, to rationalize the president’s inarticulateness as part of his ‘message’ as an ‘average American’ talking in a ‘demotic’ way. Trump was a lot worse.

Trump’s general tenor of communication was, to put it mildly, strange. In public debates, he was offensive bordering on obscene, bluntly insulting opponents (‘Crooked Hilary’, ‘the failing New York Times’) while using extravagant hyperboles in self-description and self-qualification – ‘great’, ‘the greatest’, ‘absolutely fabulous’, ‘beautiful’, ‘the best’, ‘the only one’ and so forth – and displaying a cavalier attitude towards facts as well as some of the defects earlier identified with George W. Bush (see figure 1).


Figure 1: Comment on Trump’s mispronunciation.

Trump’s public speech performances quickly became a favorite topic for late night show hosts such as Trevor Noah and Steven Colbert, and Trump imitators make a decent amount of money dissecting his usage of self-coined terms such as ‘bigly’, ‘stable genius’ and so forth and by poking fun at his obvious but stubbornly repeated gaffes (e.g. claiming that hurricane Dorian would strike Alabama, or announcing a border wall between Mexico and Colorado).

But Trump did not just talk: he also wrote a lot, and did so on Twitter. Trump’s campaign, as we know, was the first major algorithmic campaign in US history (Maly 2016), and Jordan Hollinger (2018) calls his victory the ‘first Twitter-based presidency’. His usage of Twitter is what makes his presidency entirely exceptional: he systematically used his private Twitter account as the channel for his messages, even after becoming president. The official Twitter account of the US president (@POTUS) often merely retweets messages launched by Trump on @realDonaldTrump. These tweets, consequently, fully maintain the character of ‘normal’, ‘authentic’, undoctored and unfltered tweets produced by an ‘ordinary’ Twitter user. Tweetbinder, an online repository on Trump’s tweets, claims that the president sent out about 10 tweets per day since his election, amounting to many thousands of tweets throughout his term in office. The same source also asserts that Trump writes and sends his tweets himself without the assistance (or censorship) of a communications team.2

The most amazing aspect of Trump’s usage of Twitter is the tension between his tenor as an ‘ordinary’ user of social media on the one hand, and the nature and content of his messages. Trump doesn’t just lambast his opponents or showcases his public success on Twitter, he also uses the medium to announce major (and often not otherwise announced or anticipated) policy decisions and initiatives – often causing confusion and déconfiture among his collaborators and political allies as well as drawing fierce criticism from his opponents. Twitter really is Trump’s most prominent channel of communication.

I need to pause here and turn to the general structure of communication on Twitter. And I shall start from something which all of us have absorbed during our first year of language studies: Saussure’s sender-receiver model of communication (Saussure 1960: 27). (See Figure 2)


Figure 2: Saussure’s model of communication

We see two (male) humans, A and B; A produces an utterance originating in his brain and transmits it through his mouth to the ears of B, who processes it in his brain and responds to it. All of this is very well-known, but we should remind ourselves that this simply dyadic sender-receiver model is, to a large extent, still the default model for imagining communication at large, and thus serves as the backdrop for communication theorizing. With this in mind, let us turn to the main structure of communication on Twitter. (See figure 3)

twitter schema

Figure 3: Communication structure on Twitter

We see a very different and much more complex structure of communication here. The tweet, produced by someone like Trump, is sent to an algorithm – a nonhuman ‘receiver’, if you wish – through which artificial intelligence operations forward it to numerous specific audiences (A 1, 2, …n in figure 3), whose responses are fed back, as data, to the algorithm and thence to the sender of the tweet in nonstop sequences of interaction. Parts of these audiences can relay their own uptake of the tweet (via the Twitter algorithm) to secondary audiences (A 5, 6 … n in the scheme), who can do the same – and so on, enabling a tweet to reach audiences not initially accessible. The audiences (also often called ‘bubbles’) are constructed out of users’ data yielding profiles, and they are selected on the basis of topic keywords, hashtags and histories of prior interactions.3 They consist of individuals, sure; but in the case of Trump and many other high-profile accounts also of bots – computer programs behaving like ‘normal’ Twitter users and generating specific forms of response such as liking and retweeting and sometimes dramatically increasing the volume of traffic for tweets.4

What we need to take along here is this:

(a) There is no linear sender-response structure on Twitter, because the platform itself provides an algorithmic mediator for all and any interaction;

(b) the participants are, consequently, not all human, as very crucial parts of the communication structure are controlled by automated AI technologies;

(c) as an effect of these algorithmic mediations, there is not a single ‘audience’ (or ‘public’) in the structure of communication, but a fragmented complex of ‘niched’ audiences often with incompatible interests or political orientations;5

(d) the entire system is permanently in motion, with constant interactional conversions of actions performed by (human and nonhuman) participants into data further shaping and regulating the effects of the actions (cf. Maly 2018).

We can now turn to Donald Trump’s tweets again.

Trump’s viral errors and sociolinguistic restratification

We saw how Trump’s speech idiosyncrasies were targeted by critics; his tweets have been an even more outspoken object of language-normative criticism. Given the ‘authentic’ nature of Trump’s tweets, peculiarities of writing habits can be noticed. One remarkable peculiarity is his unwarranted use of capitals – see ‘Endless Wars’ and ‘Walls’ in figure 4.


Figure 4: unwarranted capitals

The same ‘authentic’ nature of Trump’s tweets causes rather frequent typographic errors, and these are instantly singled out for condemnation. (See figure 5)


Figure 5: ‘honored’

We see indexicality in its purest form here: a typographic error leads to a judgment of the entire person: Trump doesn’t know what ‘honor’ is, hence he cannot write the word correctly. This form of sarcastic indexical interpretation is very frequent on Twitter. (See figure 6)


Figure 6: ‘passed, not past’

Those are moral condemnations of the person Donald Trump. But they are informed by something bigger: the strong cultural assumption mentioned earlier, in which we expect our social, cultural, intellectual and political elites to communicate in accordance with the most elevated standards of language – and in particular, of literate language (cf. Lillis 2013; Turner 2018). Thus, orthographic errors on Twitter are converted into judgments of Trump as president – since the president of the US is supposed to write correctly. (See figure 7)


Figure 7: ‘unpresidented’

It is because Trump is president that the indexical correctness issue is applied to his writing with such vigor and intensity. Interestingly, in such exposures, Trump’s Twitter literacy is generalized to include all of his literacy. Thus, when Trump wrote a widely publicized official letter to Turkey’s president Erdogan in October 2019, the awkward wording of the letter was caricatured by online artist El Elegante as a sequence of emojis (figure 8).

el elegante

Figure 8: El Elegante’s caricature of Trump’s letter

Twitter is the main forum for such critical exposure of Trump’s typographical errors, but it is not the only one. Mainstream media comment on them, newspapers devote articles to them, and a wide range of analysts examine them. Blogger-analyst Ginny Hogan (2018) provides a short, sarcastic summary of the problem:

“Unfortunately, the data set doesn’t include all deleted tweets, although I would be honered to learn how some of Trump’s interesting spelling choices affect tweet popularity. To bad there’s not a lot of press covfefe on that — it’s really an unpresidented phenomenon #Denmakr.”

The reference to ‘covfeve’ here is interesting, because it’s probably Trump’s most iconic Twitter error. Trump posted it in May 2017, and the nonsense word is probably a botched attempt to write the term ‘coverage’ (see figure 9).


Figure 9: ‘covfeve’

The word became an instant hit among critics on Twitter and beyond, the more since the White House Press Secretary tried to explain it as meaningful: “I think the president and a small group of people know exactly what he meant”, Sean Spicer announced.6 ‘Covfeve’ became the stuff of memes and went viral in a wild stampede of (often hilarious) critical uptake.

So far so good: we see how orthographic errors by Donald Trump lead to relatively predictable – standard – indexical interpretations as transgressive and inadmissible features of communicative conduct displayed by the president of the United States. We can observe the dominant sociolinguistic stratification at work here: such errors in writing are wrong, certainly when performed by members of the elites, and they index moral disqualification of the person and question his membership of those elites. Someone who commits such errors should never be president of the US, is the line of interpretation we have observed so far. And this would be the end of the story in Saussure’s communication model: B (the audience) has disqualified what A (Trump) tried to communicate. But as we have seen, communication on Twitter is different.

Let us have a look at the people who posted the critical comments on Trump’s errors. All of them are public figures: Noga Tarnopolsky is a journalist, RC de Winter is a poet and digital artist, El Elegante is a digital artist, Randy Mayem Singer is a successful movie and TV series screenwriter, and J.K. Rowling is of course the author of the Harry Potter blockbusters. All of them are intellectuals and artists working with language, and in the worldview of Donald Trump and his supporters, they belong to the (‘liberal’) cultural ‘elites’. Within those ‘elites’ they form a subgroup notoriously critical of Trump and his politics, and Trump himself takes shots at such liberal intellectual and artist elite figures quite often on his Twitter account. (See figure 10)


Figure 10: Meryl Streep is over-rated.

These intellectual and artistic elites clearly form one (or several) of the niche audiences on Trump’s Twitter account – a hostile one. And they can be described, by the Trump camp, as the elites whom Trump wants to defy and defeat, for they are in opposition to ‘the people’. Many actors in Trump’s universe are ‘a threat/enemy to the people’ – mainstream media are, for instance, quite systematically qualified as such.7 Ridiculing Trump’s orthographic errors (or speech habits) can thus be represented as a predictable and stale anti-Trump reaction coming from one of the elite social groups he targets as opposed to the interests of ‘ordinary Americans’.

This is the point where we get sociolinguistic restratification. Trump’s orthographic errors are (very much like George W. Bushes discursive inarticulateness) indexically upgraded from ‘bad in the eyes of the elites’ to ‘good in the eyes of the people’ – they become indexically restratified as the demotic code that iconicizes the down-to-earthness of ordinary Americans. And this restratified sign goes viral among the other and more supportive audiences of his Twitter account. In figure 11, we see how a Trump supporter uses #covfeve (followed by two positive emojis) as an emblem of pride used against Trump critics. The meaning attributed to the word here is grounded in the interpretation of Trump’s initial ‘covfeve’ tweet, which attacked mainstream media. This intertext provides the function of the word here: covfeve has become (like ‘MAGA’) a term that can be used to talk back to Trump’s detractors.

protrump blanked

Figure 11: pro-Trump Twitter account.

The term ‘covfeve’ was also adopted by a score of Twitter users in their user names. (see figure 12)


Figure 12: ‘covfeve’ accounts

Some of these accounts are obviously held by people who are critical of Trump, while others are held by Trump supporters. The indexical vectors of the term are opposites: for pro-Trump people, ‘covfeve’ indexes support for Trump and hostility towards his elite critics; for anti-Trump people, it indexes the fact that Trump is unfit for the presidency. And both indexical vectors are attached to an orthographic error made on a public forum such as Twitter. ‘Covfeve’ became a viral error, circulated within very different audiences and with very different meanings.

A lab of restratification

Let me summarize the case. Trump’s orthographic errors on Twitter got immense traction on Twitter (and beyond) and did so within very different audiences, some of whom applied the ‘standard’ sociolinguistic stratification in which orthographic correctness is mandatory for people at the top of the social ladder. Other audiences used an entirely different, ‘demotic’ understanding of these errors, presented there as emblematic of someone intent on defending the interests of ‘ordinary’ Americans. The virality of errors such as ‘covfeve’ implies at least two entirely opposite indexical vectors, one of which restratifies the conventions of the sociolinguistic domain of writing from elite-dominant to demotic-dominant.

There is, of course, irony in the fact that Donald Trump (like George W. Bush before him) can be presented at all as a non-elite, ‘ordinary’ person. He is a scion of a very wealthy family and proudly proclaims his wealth to all who want to listen, he was a mass media superstar, a bestselling author and an alumnus of the University of Pennsylvania’s prestigious Wharton School, and he is of course the president of the United States. From what is publicly known about his lifestyle, he really doesn’t live like ‘ordinary’ Americans.

His communication styles, however, offer the potential to turn this obvious misfit into a perfect fit: sarcasm about his speaking and writing errors can be presented as ‘elitist’ and magnified – generalized – as part of a pattern of elite domination of ‘ordinary’ Americans, the kind of elite domination Trump promised to abolish as president. In the process, the sociolinguistic norms of different audiences are played off against each other in Twitter discussions. It is on Twitter that the fragmented nature of audiences affords us a glimpse of the fragmentation of sociolinguistic stratification, with ‘standard’ (i.e. ‘elite’) norms competing with demotic ones. Within the latter, errors are not just normal or acceptable, they are prestigious and emblematic, as we could see in figure 11. The errors are there for a good reason: they iconicize the perceived ‘big’ divisions in US society and the perceived exclusion of ‘ordinary’ people from major public debates. Trump’s errors are icons of the voice of such ‘ordinary people.

We see a complex, polycentric sociolinguistic system here, in which specific norms can dominate specific segments of the public domain while they are being fundamentally challenged in other segments. Social media such as Twitter make this polycentricity and its restratifying features abundantly clear: they are a veritable lab for examining sociolinguistic normativity, debates and contests about normativity, and innovations in that field (cf. Blommaert 2018; Seargeant 2019).

For sociolinguistics as a science, this means that the supposed stability of stratified sociolinguistic systems – with minorities and majorities clearly demarcated by lines of objective power – needs to be critically revisited, empirically as well as theoretically. In the online-offline nexus, heteronormativity is not an exception, but a rule among segments of the users’ communities. These segments now have acquired public channels of communication, making previously invisible and disqualified demotic forms of language and literacy available for uptake, and turning them into prestige-carrying varieties demanding respect and public recognition. This new politics of language is expertly used by politicians such as Trump as well as by other powerful political and economic actors: the play of stratification and restratification is at the heart of several very large processes of social change, and requires a sociolinguistic analysis that does justice to its complexity.


  1. I am dedicating this essay to my friend and colleague Sjaak Kroon, with whom I collaborated intensely for over a decade and with whom I discussed almost any idea that came into being during that time. I tailored the essay in such a way that it addresses several of Sjaak’s interests, overlapping with mine. I am grateful to Ico Maly for critical comments and suggestions on an earlier version of the paper.
  2. See On the Trump Twitter Archive, an almost comprehensive collection of Trump’s tweets can be found. See As for Tweetbinder’s claim that Trump is the sole author of his tweets: I afford myself some doubt. Surely, he is the author of a huge number of tweets, but there are stylistic differences between his tweets (a full analysis of which is reserved for another paper) that point towards more hands touching his Twitter keyboard.
  3. Hogan (2018) provides some insights into the traction profile of Trump’s Twitter account. We should remember that there is another, human filter on what is being shown on social media such as Facebook and Twitter: the platform guidelines and restrictions on content, prohibiting, for instance, explicit sexual content, hate speech or violent images to be publicly visible, and policed by (often subcontracted) individuals. The criteria applied, along with the practices, outcomes and labor conditions in this domain are the object of constant controversy. See Varis (2018) for a discussion.
  4. In late October 2019, Donald Trump’s Twitter account boasted over 66 million followers. But the @realDonaldTrump account has been shown to contain an unusually large number of bots among its followers. See For the effects of bots on the intensity of Trump’s Twitter traffic, see
  5. To clarify the heterogeneity of Trump’s audiences: given the sheer importance of his tweets as political statements and announcements, his Twitter community is not necessarily made up of ‘followers’ in the sense of people who agree with or support Mr. Trump. Reporters and opponents are also compelled to follow his account in order to stay abreast of what the president has in mind.
  6. For a retrospective report, see
  7. For a recent critical review of Trump’s ‘enemy of the people’ argument, see


Agha, Asif (2007) Language and Social Relations. Cambridge: Cambridge University Press

Arnaut, Karel, Jan Blommaert, Ben Rampton& Massimiliano Spotti (eds) (2016) Language and Superdiversity. New York: Routledge

Bauman, Richard & Charles Briggs (2003) Language and Modernity: Language ideologies and the politics of inequality. Cambridge: Cambridge University Press.

Bernstein, Basil (1971) Class, Codes and Control, Vol 1: Theoretical studies towards a sociology of language. London: Routledge & Kegan Paul.

Blommaert, Jan (2005) Discourse: A Critical Introduction. Cambridge: Cambridge University Press

Blommaert, Jan (2008) Grassroots Literacy: Writing, identity and voice in Central Africa. London: Routledge

Blommaert, Jan (2010) The Sociolinguistics of Globalization. Cambridge: Cambridge University Press

Blommaert, Jan (2015) Chronotopes, scales and complexity in the study of language in society. Annual Review of Anthropology 44: 105-116

Blommaert, Jan (2016) From mobility to complexity in sociolinguistic theory and method. In Nikolas Coupland (ed.) Sociolinguistics: Theoretical Debates: 242-259. Cambridge: Cambridge University Press.

Blommaert, Jan (2018) Durkheim and the Internet: On Sociolinguistics and the Sociological Imagination. London: Bloomsbury.

Blommaert, Jan (2020) Formatting online actions: #justsaying on Twitter. In Jerry Won Lee & Sender Dovchin (eds.) Translinguistics: Negotiating innovation and ordinariness: 75-89. London: Routledge

Boas, Franz (1911) Introduction. Handbook of American Indian Languages, Vol. 1:  1-83. Bureau of American Ethnology, Bulletin 40. Washington: Government Print Office (Smithsonian Institution, Bureau of American Ethnology).

Boas, Franz (1928) Anthropology and Modern Life. New York: W.W. Norton & Company

Bourdieu, Pierre (1987) Distinction: A social critique of the judgment of taste. Cambridge MA: Harvard University Press.

Bourdieu, Pierre (1991) Language and Symbolic Power. Cambridge: Polity

Darnell, Regna (1998) And Along Came Boas: Continuity and revolution in American anthropology. Amsterdam: John Benjamins

Eastman, Carol (1983) Language Planning: An Introduction. San Francisco: Chandler & Sharp

Fabian, Johannes (1986a) Language on the Road: Notes on Swahili in two Nineteenth-Century travelogues. Hamburg: Buske Verlag.

Fabian, Johannes (1986b) Language and Colonial Power: The appropriation of Swahili in the former Belgian Congo 1880-1938. Berkeley: University of California Press.

Fishman, Joshua (1971) Bilingualism in the Barrio. Bloomington: Indiana University Press.

Gumperz, John (1982) Discourse Strategies. Cambridge: Cambridge University Press

Hogan, Ginny (2018) Twitter advice for President Trump: A statistical adventure. Blogpost,

Hollinger, Jordan (2018) Trump, social media, and the first Twitter-based presidency. Diggit Magazine 7 May 2018.

Hymes, Dell (1980) Language in Education: Ethnolinguistic essays. Washington DC: Center for Applied Linguistics.

Hymes, Dell (1983) In Vain I Tried to Tell You: Essays in Native American Ethnopoetics. Philadelphia: University of Pennsylvania Press

Hymes, Dell (1992) The concept of communicative competence revisited. In Martin Pütz (ed.) Thirty Years of Linguistic Evolution: 31-57. Amsterdam: John Benjamins

Hymes, Dell (1996) Ethnography, Linguistics, Narrative Inequality: Toward an understanding of voice. London: Taylor and Francis.

Kroon, Sjaak & Jos Swanenberg (eds.) (2019) Chronotopic Identity Work: Sociolinguistic Analyses of Cultural and Linguistic Phenomena in Time and Space. Bristol: Multilingual Matters.

Kroskrity, Paul (ed.) (2000) Regimes of Language. Santa Fe: SAR Press.

Labov, William (1970) The logic of nonstandard English. In Frederick Williams (ed.) Language and Poverty: Perspectives on a theme: 153-189. New York: Academic Press

Lempert, Michael & Michael Silverstein (2012) Creatures of Politics: Media, message, and the American Presidency. Bloomington: Indiana University Press.

Lillis, Theresa (2013) The Sociolinguistics of Writing. Edinburgh: Edinburgh University Press.

Maly, Ico (2016) How did Trump get this far? Diggit Magazine 17 October 2016.

Maly, Ico (2018) Algorithmic populism and algorithmic activism. Diggit Magazine 8 November 2018.

Myers-Scotton, Carol  (1993) Social Motivations for Codeswitching: Evidence from Africa. Oxford: Clarendon Press.

Phillipson, Robert b(1992) Linguistic Imperialism. Oxford: Oxford University Press.

Sapir, Edward (1921) Language. New York: Harcourt Brace & Company

Saussure, Ferdinand de (1960) Cours de Linguistique Générale (eds. Charles Bally & Albert Sechehaye). Paris: Payot.

Silverstein, Michael (1979) (1979) Language structure and linguistic ideology. In Peter Clyne, William Hanks & Carol Hofbauer (eds.) The Elements: A Parasession on Linguistic Units and Levels: 193-247. Chicago: Chicago Linguistic Society.

Silverstein, Michael  (1996) Monoglot ‘standard’ in America: standardization and metaphors of linguistic hegemony. In Don Brenneis & Ronald Macaulay (eds.) The matrix of language: Contemporary linguistic anthropology: 284-306. Boulder: Westview Press.

Silverstein, Michael  (1998) Contemporary transformations of local linguistic communities. Annual Review of Anthropology 27: 401-426.

Silverstein, Michael (2003) Talking Politics: The substance of style from Abe to ‘W’. Chicago: Prickly Paradigm Press.

Seargeant, Philip (2019) The Emoji Revolution: How technology is shaping the future of communication. Cambridge: Cambridge University Press

Turner, Joan (2018) On Writtenness: The cultural politics of academic writing. London: Bloomsbury.

Varis, Piia (2018) Labouring in the digital economy: The people making content (in)visible online. Diggit Magazine 1 November 2018.

Whorf, Benjamin Lee (1956) Language, Thought and Reality: Selected Writings by Benjamin Lee Whorf (ed. John B. Carroll). Cambridge MA: MIT Press.


Bullshit: the struggle goes on

language: a feminist guide

When it comes to the way she speaks, a woman’s place is in the wrong. It’s a point I’ve made frequently on this blog, and last week brought a reminder of how true it continues to be. On the same day I published a post inspired by criticisms of Greta Thunberg’s ‘strident’ speech to the UN Climate Action summit (‘strident’ being a code-word for women who express their views in an ‘excessively and unpleasantly forceful way’), the Times Education Supplement published a piece complaining that women don’t speak forcefully enough. It started like this:

“I’m sorry, I’m no expert on this but could we possibly…”

Have you ever heard yourself say something like this to your team?

I certainly have and I cringe to think how I must have come across.

Such phrases can make women appear weak or ineffective to colleagues, which in turn may affect whether they…

View original post 2,016 more words

Constructing and consolidating chronotopes

A man dressed in traditional Ukrainian Cossack clothes smokes in a phonebooth during a rally to support EU integration in Kiev

Jan Blommaert

(Discussant’s comments, panel on Mobility, marginality and meaning: A chronotopic approach, convenors Lydia Catedral and Farzad Karimzad. IPrA Conference, Hong Kong, June 2019)

There are several things that I find extraordinarily inspiring about this panel and in what follows, I shall unfortunately have to restrict myself to comments on a small selection of themes only. I will formulate my comments from the position of a participant in an ongoing conversation, and not from that of a critical reviewer. The latter position would presuppose an authority I cannot, and do not wish to claim for myself. Even if many papers in the panel draw on aspects of my work, that work is merely an input in far more productive and sophisticated exercises than the ones I was able to offer. I am deeply grateful to the panel organizers and its presenters for having the opportunity to engage with a wave of very advanced work on topics that are central to my own concerns. My comments, consequently, will reflect things that I have learned from the presentations in this panel – insights I find valuable for my own development.


I must start at the most trivial level, by observing that conceptual work on chronotopes is unfinished but that several important developments can be noted. The fact that work is unfinished is translated in a degree of messiness in the use of chronotope as a notion in the various papers. There are authors who seek refinement to the notion by returning to the canonical formulations of Bakhtin and consequently arrive at a more restricted, somewhat ‘closed’ analytical notion of chronotope referring to invokable historicities (e.g. Rampton & Sankaran, Rowlett & King); there are authors who use it as a more open and flexible heuristic notion, capable of being tested as a tool for enriching existing approaches (e.g. Bhatt); and in need of further conceptual refinement (e.g. Karimzad, Bolonyai). In the latter papers, chronotope is used as a potentially productive and more precise gloss for what Goffman called ‘the situation’ (a thing we tended to neglect, as Goffman explained).

Both directions are useful and compatible – under certain conditions to which I shall turn in a moment. For now, let me briefly refer to what Goffman actually wrote on ‘the situation’.

“A student interested in the properties of speech may find himself having to look at the physical setting in which the speaker performs his gestures, simply because you cannot describe a gesture fully without reference to the extra-bodily in which it occurs. And someone interested in the linguistic correlates of social structure may find that he must attend to the social occasion when someone of given social attributes makes his appearance before others. Both kinds of students must therefore look at what we vaguely call the social situation. And that is what has been neglected.” (Goffman 1964: 134)

Goffman connects two elements here, both of which appear as compelling contextual factors in analysis. First, there is the “physical setting” within which interaction occurs – the actual timespace constellation within which people encounter each other, in other words. This timespace constellation, we can add, is infrastructural and, thus, material – a point often clearly ‘neglected’ (in Goffman’s terms), but inevitable as a feature of online communication where the sociotechnological infrastructures for communication pre-inscribe chronotopic affordances for their users (see the papers by Prochazka and Lyons, Tagg & Hu).

Goffman adds to this a second element: “the social occasion”. The latter is defined (with an oblique reference to Durkheim’s “social fact”) as “a reality sui generis” within any social system, and it stands for the rules of participation and communicative behavior that provide “scripts” (if you wish) ordering concrete communicative events between people who carry “given social attributes”. Both elements – note – are coordinated in actual interactional events. It is this dialectic of mutual influences between settings and social scripts that shapes the “joint social orientation” characterizing social interaction, which enables Goffman (id: 135) to provide his own, interactional, definition of the social situation:

“I would define a social situation as an environment of mutual monitoring possibilities, anywhere within which an individual will find himself accessible to the naked senses of all others who are “present” and similarly find them accessible to him.”

As we know, much of Goffman’s work was focused on the precise description of specific social situations – think of the poker game in Encounters (1961) and the lecture in Forms of Talk (1981) In each of these situations, Goffman emphatically pointed to the ways in which situations came with sets of conditions on participation, rules of engagement and forms of communicative action. Concrete and socioculturally recognizable timespace configurations involve nonrandom modes of social action and lead to specific social effects – that is the major insight we can get from Goffman’s oeuvre, and which resonates with the work of scholars inscribed in the same lines of inquiry (think of Garfinkel, Cicourel, the Goodwins and the Scollons). It is this insight for which chronotopes can be a helpful gloss.


So there is a messiness to chronotope in the panel. And that is fine, for it points towards an older problem for which new solutions are being explored. That older problem is the problem of ‘context’; or more precisely, the way in which problems of contexts have tended to be subdivided into issues about ‘what is context’ and issues about ‘what happens in context’. Chronotope, I believe, compels us to merge both issues. And now I must engage with some of the conditions for compatibility to which I hinted earlier.

The conceptual messiness in the panel should not obscure some extremely useful points of agreement across different papers, articulated explicitly in some and implicitly in others.

  1. A first, and major point of agreement I observe is: chronotopes are not just there as a priori ‘structures’ which one can walk into and out; they are made by participants in concrete social action. Karimzad used ‘chronotopization’ as a term reflecting this action-centered perspective. In order to illustrate it, we can go back to one of Garfinkel’s (2002) classic examples: the queue. A queue is (and few would dispute this) a chronotope. Garfinkel describes the ways in which people form queues in a seemingly unprompted way as congregational work (social work done collectively, for no reason other than the fact that there is a collective present or in formation) in which recognizable forms of social order are constructed – a ‘format’ as he calls it, with ‘autochtonous order properties’ i.e. “empirically observable properties of the congregational work of producing social facts” (2002: 245). Converted into the vocabulary of ‘context’, we see how Garfinkel (like many others) views contexts as evolving out of social action. But he does not stop there. As soon as a queue has been formed, it becomes an extremely normative and policed timespace configuration. The ‘personnel’ in the queue have a very strong awareness of the order it embodies, and will react strongly whenever people attempt to violate that order (e.g. by jumping the queue or abusing the physical proximity in the queue for transgressive talk or conduct). The queue generates a temporary (but quite robust) community tied together by the ‘authochtonous order properties’ of the queue. Returning now to the main point here, what we see is that contexts are evolving out of action, and consolidated through action as well. What context is and what happens in context can no longer be separated – they have merged. (We see excellent examples of this dynamic of context construction and consolidation in Prochazka’s paper.)
  2. If we adopt this action perspective, let’s look more closely at the kinds of actions people perform. In the data shown in all the papers (Lyons, Tagg & Hu; Prochazka; Rowlett & King; Karimzad; Choi & Lo; Djuraeva & Catedral; Bolonyai; Bhatt; Rampton & Sankaran) we see that chronotopic congregational work is argumentative, it is done by people in order to make a point. It is, when seen from these examples, a resource for constructing ‘facts’, telling ‘the truth’ and convincing others of this truth. The how of such argumentative usage depends on what specific chronotopes afford – it is part of the work of consolidation, we can say. The what of such usages are specific forms and bits of invokable historicities deployed in specific discursive-argumentative actions. I think there might be an area here of particular interest for further and more detailed research, and my final two points will offer some suggestions for that.
  3. It is entirely possible that these specific forms of argumentative usage explain, at least partly, some other things about which several presenters in this panel appear to agree. There is such agreement, for instance, on the non-unified nature of chronotopes – their scaled character; on their unstable, unfinished and porous character observable through cross-chronotopic shifts and connections and through the occurrence of ‘chronotopes within chronotopes’; on the particular ways in which ‘big’ chronotopes such as those of Modernity and colonialism (or of contemporary globalization and nationalism) pervade and organize ‘small’ situational chronotopes of semiotic deployment (much in the way of Goffman’s frames, I would suggest); and how narrative voices from the margins inevitably orient towards polycentric and scaled chronotopes; and so forth. I used the term ‘synchronization’ as a way to capture how such complexly layered forms of semiosis collapsed into (chronotopically) situated moments of performance and uptake. I would now add the suggestion that we see such phenomena as part of particular forms of argumentation.
  4. Another dimension might evolve more clearly from such an exercise. And here too, I can refer to something that struck me in almost all the papers when I was reviewing the data used. They were replete with moralizations. Now, of course we know that there is something inherently moral about ‘knowledge’, ‘facts’ and ‘the truth’ (the Goodwins never stopped reminding us of that fact). But what we can see now, is how the argumentative work done in the construction and consolidation of chronotopes strongly revolves around making legitimate points – references to war and genocide, of displacement, traditions, the general gender, race and class relations prevailing in specific social environments, are offered as the moral truth that others should be persuaded by. We find ourselves here in a different metalinguistic realm: one in which the deployment of specific semiotic resources indexes fundamental moral stances blended with epistemic and affective ones and with identity projections. Here, too, I anticipate very stimulating areas for further inquiry.


Which takes me to my conclusion. The comments I have made are testimony to the exceptionally creative and inspiring work done in this panel. I do believe that the work presented here takes our field forward, and that it does so in a way which is increasingly disfavored in science: by exploring, unthinking, reimagining, testing and toying – the fundamental work of coming up with ideas which all of us need so badly in order to re-search, to search again for what we think we already know. To the panel organizers and presenters: mes hommages.


Garfinkel, Harold (2002) Ethnomethodology’s Program: Working Out Durkheim’s Aphorism (ed. Anne Warfield Rawls). Lanham: Rowman & Littlefield.

Goffman, Erving (1961), Encounters: Two Studies in the Sociology of Interaction, New York: Bobbs-Merrill.

Goffman, Erving (1964) The neglected situation. American Anthropologist 66/6 (part 2): 133-136.

Goffman, Erving (1981), Forms of Talk, Philadelphia: University of Pennsylvania Press

Sociolinguistic scales in retrospect

0024 copy

Jan Blommaert

Productive ambivalence

The idea of sociolinguistic scales points towards the non-unified nature of sociolinguistic phenomena, and it contains a productive ambivalence. It refers to what we could call scope of communicability, which is, if you wish, a horizontal image of spread, dimension, degree; but it refers simultaneously to value, distinction, quality in the sociolinguistic field, which is a vertical image of stratification. This is the ambivalence. What ‘scale’ does, is to express an intersection of scope and value, and this is why the ambivalence is productive. It generates a heuristic in which we interrogate sociolinguistic facts as simultaneously scoped and valued, subject to forces and effects that have to do with issues of resource distribution, availability and access, as well as with issues of sociocultural and political uptake – of size and importance, of quantity and quality, one can say. And distinctions observable along those two axes explain the dynamics of sociolinguistic life and the main features of concrete sociolinguistic economies. Every difference in scope is likely to be accompanied by a difference in value, and this essential non-neutrality of sociolinguistic resources (in the broadest sense of the term) is at the heart of concrete meaning-making practices.

Scales have become a topic of considerable interest in the last decade among scholars of language-in-society. It is a conceptual cornerstone of any sociolinguistics of globalization, and it has helped scholars reimagine crucial notions such as ‘context’. The present collection of essays is testimony to the ongoing productivity of the concept and to the many ways in which it enables creative analysis, precisely articulating the ambivalence I sketched above: analysis of scope of communicability, of the ‘reach’ of resources, technologies and infrastructures, joined with analysis of metalinguistic – sociocultural and political – valuation and its effects on the attribution of meaning and identity in interaction. Many scholars (and I include myself) currently view scales as an indispensable concept, but articulate an uneasiness about it – it is precise and clumsy at the same time, it creates useful lines of argument but instantly complicates such arguments, and while it has an immediate resonance of transparency (we know what it indicates), it also is hard to define with some precision.

In what follows, I will suggest that we need a concept of sociolinguistic scales, but we also need more than that; and I will suggest that we need scales as an imaginative concept, a loosely descriptive one that generates a number of other issues, to be addressed with more precise tools. I shall do so by reviewing my own motives for using sociolinguistic scales.

Three problems

In the early 2000s, the topic that consumed most of my energy was that of voice and globalization. I was intensively working on problems of communication in asylum applications, mainly by asylum seekers from Africa in Western Europe. Gradually, I began to identify the core issue as one of resources and mobility. African asylum seekers brought sociolinguistic resources to the procedural interviews in Europe; and while such resources – think of African varieties of English or French – were perfectly adequate in Africa, they were very often dismissed as inadequate in Europe and their users remained voiceless in the asylum procedure. It appeared that in globalization contexts such as those of asylum, some resources afforded mobile voice, in the sense that they were valued as meaningful across entirely different timespaces, while others didn’t – something that could be reformulated in terms of different scale affordances, differences at the intersection of scope and value.

This insight – the scaled character of sociolinguistic phenomena in contexts of globalization – enabled me to start addressing three substantive problems:

  1. The problem of meaning: what goes on in actual meaning-making processes?
  2. The problem of situatedness, or more broadly formulated, of context.
  3. The problem of indexicality: a logical extension of the two first problems given the locus of indexicality in text-context relationships.

Let me turn to these problems now and indicate how sociolinguistic scales, as an imaginative concept, enabled me to come up with solutions for these problems.

The problem of meaning

The work on asylum seekers became the empirical basis for a book called Discourse (2005). The book offered a theory of sociolinguistic inequality related to voice and globalization, the core of which was a view of meaning as layered and composed of different historically loaded elements (from slow-and-big things such as ‘English’ to infinitely small aspects of interactional deployment), synchronized in interaction.

I glossed this view as ‘layered simultaneity’ and it merits some explanation. The point to it is that whenever we communicate, we deploy a variety of intertextual resources, such as topics, interlocutors, language, accent, technologies, registers and so forth. While all of these resources will be simultaneously deployed, we must realize that they are of a different order – here comes scale. English, for instance, is of a different order in interactions than the actual interlocutor with whom we interact in English or the topic we address with that interlocutor in English (English transcends the specificity of interlocutor and topic). And these different orders will generate different effects in communication: shifting from English into Swahili, for instance, will exclude more interlocutors than switching from a Cockney into an RP accent in English. So scope and value walk hand in hand: the value of resources in interaction is often tied to their scope of deployment, to their affordance to include or exclude interlocutors, topics and communication practices. I elaborated the latter issue in Grassroots Literacy (2008).

The concept of scales enabled me here to imagine meaning as non-unified and stratified, as an object the different layers of which could not be imagined as equivalent, even if they co-occurred in moments of synchronization. Such synchronizations had to be seen as complex processes in which very different forces co-occur, and in which the differences between such forces mattered very much in view of the outcome of communication. But this was merely one step in a longer argument, just one of several problems solved, and more was needed.

The problem of situatedness

Meaning emerges in contexts – actual, concretely situated contexts. And context, as we know, is a notoriously elusive notion in analysis. I was deeply influenced by Aaron Cicourel’s views of context (1967; 1992) as multiple, layered and stacked, and access to contexts as unevenly distributed – a medical doctor has access to contextual resources not accessible for the patient, very much like security services having access to contexts not open to inspection by their suspects. Scale once more provided a useful prism through which we could contemplate such non-unified contexts.

The fundamental given of non-unified contexts, of course, called into question what we mean by the notion of ‘the local’ in the analysis of interaction. ‘The local’ (as in ‘locally performed’) is often used as an equivalent for the qualification of ‘situated’ when we decode communicative events: a synchronized here-and-now that operates as a self-contained and self-sufficient reservoir of inferential meaning (codified, for instance, in Conversation Analysis in the Schegloffian tradition, see Blommaert 2001). From the work on asylum seekers, I had understood that these self-contained and self-sufficient dimensions of situatedness made no sense, for we were continually confronted with an interplay of ‘local’ and ‘translocal’ features of contexts of globalized mobility, and with uniquely relevant and situationally contingent inferences operating alongside generic inferences. Narrative patterns performed by African asylum seekers clashed with institutional expectations about narrativity in a European bureaucratic and forensic tradition; they did so situationally of course, an each time in partly unique ways; but the uniqueness of such cases was overrun by generic differences in production and uptake of narratives (something also documented with respect to written stories in Blommaert 2008). So we encounter two kinds of situatedness synchronically operating in concrete events: unique situatedness as well as generic situatedness – the latter containing actualizations of the genres, frames, formats that generate the moralized behavioral scripts for meaning-making we usually qualify as ‘cultural’ (cf. Blommaert 2018). It is useful to underscore that both dimensions of situatedness are of a different order.

Which is why a further step was needed at this point. The layering of contexts-for-inferencing entailed, in actual moments of deployment, the layering of norms. Norms, too, are non-unified and scaled phenomena in communication, and we must assume that (given the layering) that at any moment of communication, different sets of norms are present in the situation, and all of them can be invoked by participants. I called this layered-normative dimension of communication ‘polycentricity’, and this concept became, along with scales, one of the key terms in The Sociolinguistics of Globalization (2010). And norms take us straight into the third problem.

The problem of indexicality

Given the treatment of the two previous problems, the third one is relatively simple. Indexicality stands for the meaning effects generated from text-context relationships, and if texts as well as contexts are non-unified, we need to make similar distinctions in the field of indexicality. In a polycentric situation, thus, various differently scoped and valued orders of indexicality are simultaneously at play, as emic regulators of meaning-making. Orders of indexicality – a concept obviously influenced by Foucault’s ‘orders of discourse’ – was the notion that completed the conceptual triad of The Sociolinguistics of Globalization, alongside scales and polycentricity. What I intended to express through that notion were general forms of ‘normalcy’ in social interaction, the available (but not evenly accessible) scripts for ‘normal’ meaning-making in situated communicative events.

As I said, such orders of indexicality are deployed as emic regulators of meaning-making; in less convoluted terms, they are the patterns of communicative conduct that generate recognizability in interaction: particular actions are recognizable as something specific, as, for instance, a joke, a lecture, flirting, a friendly response, the purchase of a railway ticket, and so forth. As in the discussions of the previous problems, we must assume that orders of indexicality are – literally – of a different order and regulate, at once, the aspects of meaning that are given and those that are new, the generic ones as well as the uniquely creative ones. The same imagery of scales applies here, and as before, it enables us to make more specific distinctions.

Always important, never by itself

In reviewing the line of argument I applied in solving the problems of meaning, situatedness and indexicality, I hope that one things has become clear. Scale was always a starting point, because it was the imagination of scaled phenomena that transformed them into objects that could be differently addressed. In my (admittedly idiosyncratic) development, scale has been extraordinarily productive as an imaginative concept, an instrument enabling a fundamental revision of what we believe certain facts to be – a tool for the creation of ideas, one could say.

Once these ideas were there, however, the scaled characteristics of sociolinguistic phenomena demanded more fine-tuned and accurate conceptual instruments and analytic strategies. The idea of sociolinguistic scales inevitably leads to a sociolinguistic phenomenology in which instability, plurality, indeterminacy and low presupposability feature – it leads to a sociolinguistics of complexity, in sum (Blommaert 2016). And for addressing complexity, notions such as scales carry too much of a suggestion of stable, static and clear-cut distinctions (as when we speak of ‘micro’ and ‘macro’ in relation to scales) and risk being too blunt an object for making the microsurgical distinctions we need to be able to make in actual analysis.

This was certainly how I experienced working with sociolinguistic scales: they were always indispensable in preparing the canvass and sketching the outlines of analysis; but once the analysis itself had to be done, scales were merely the parameters within which I operated, and the operation itself demanded a more elaborate, systematic and complex set of conceptual and methodological tools.



Blommaert, Jan (2001) Context is/as critique. Critique of Anthropology 21/1: 13-32.

Blommaert, Jan (2005) Discourse: A critical introduction. Cambridge: Cambridge University Press

Blommaert, Jan (2008) Grassroots Literacy: Writing, Identity and Voice in Central Africa. London: Routledge

Blommaert, Jan (2010) The Sociolinguistics of Globalization. Cambridge: Cambridge University Press

Blommaert, Jan (2016) From mobility to complexity in sociolinguistic theory and method. In Nikolas Coupland (ed.) Sociolinguistics: Theoretical Debates: 242-259. Cambridge: Cambridge University Press.

Blommaert, Jan (2018) Durkheim and the Internet: Sociolinguistics and the Sociological Imagination. London: Bloomsbury

Cicourel, Aaron (1967) The Social Organization of Juvenile Justice. New York: Wiley.

Cicourel, Aaron (1992) The interpenetration of communicative contexts: Examples from medical encounters. In Alessandro Duranti & Charles Goodwin (eds.) Rethinking Context: 291-310. Cambridge: Cambridge University Press.

See also this video.


When your field goes online


When your field goes online:

Ethnographic fieldwork in the online-offline nexus

Jan Blommaert & Dong Jie

(Draft postscript to Ethnographic Fieldwork: A Beginner’s Guide. Second and enlarged edition. Bristol: Multilingual Matters, in press)

When we wrote the first edition of Ethnographic Fieldwork in 2008-2010, social life was still very much seen as an offline affair. People used to refer to the digital world as the virtual one, implying that it was in some way not part of the real world. As for new media, Facebook was an infant and the iPhone was a toddler when we wrote the book, and social media activities were widely seen as a relatively irrelevant add-on of ‘real’ (read: offline) social life.

The online-offline nexus

A decade later, this can obviously no longer be maintained. The online world is now fully integrated with the offline one, in the sense that very few of our ordinary, everyday activities proceed without being in some way affected by online infrastructures; and very many of such activities can only proceed due to the existence of such online dimensions of life. From making photographs with our smartphones to checking the weather app, the traffic app, or our daily fitness routine app, and from online shopping, travel booking, banking and reading to quick searches (aptly called, in many places, “Googling”), to TV-on-demand binging, vloggers and influencers, livestreamed events and commercial as well as political campaigns waged on social media – our social, cultural, economic and political lives have changed dramatically. The widespread use of social media has transformed the media and popular culture landscapes globally and has shifted the boundaries between the private and the public spheres. And each action we perform online, however minute, generates data that are aggregated into new systems of surveillance and control and affect our lives in mostly invisible ways. Note that while such developments are spread unevenly across the globe, there are few places in the world where they are not experienced to some degree.

These phenomena are by now well documented, so we don’t think that a full survey of them is warranted here. The fundamental fact we have to take on board is: we live our lives largely in an online-offline nexus, in which both dimensions are equally vital and indispensable. Yet, when it comes to social theory and method, we still very much continue to approach these lives from within frameworks developed to describe and analyze an offline world – and ethnography is no exception to this (Kaur-Gill & Dutta 2017; Blommaert 2018; Varis & Hou 2019). This is not unusual: theory is always slow to catch up with changing realities, and theories that incorporate change as a fundamental given are few and far between. The same goes for method: scholars are usually reluctant to surrender tools of investigation of which they believe that they worked adequately in the past.

When it comes to ethnographic fieldwork, however, we cannot avoid issues of theoretical and methodical adequacy, for a very simple reason: in the online-offline nexus, the field where we do our fieldwork has gone online, and we need to follow that route if we wish to adequately address what it is we observe and analyze.

In what follows, we will offer three reflections on this new field and show how they complicate matters for ethnographers (and others). To be sure, things were complicated enough in an offline field; when we incorporate the online field, however, several new things require focused attention. We need to add some question marks to three seemingly unproblematic things: what do we see? Who is there? And where are we in an online-offline fieldwork site.

What do we see? The compelling bubble

The first complication is caused by what is known as the ‘bubble effect’: whenever we go online, we find ourselves in a space the structure and composition of which has been configured algorithmically, on the basis of data profiles for specific users, machines and software tools. And this is an absolute given: there is no actual PC or smartphone in the world that offers its user an unrestricted view of the online world. Not to put too fine a point of it: whenever you go online on any device, anywhere and anytime, you will encounter bias, and there is simply no neutral and unbiased position of observation possible in the online world. This is worth remembering: using the PC of, say, your local public library to do online research doesn’t remove the bubble effect. It merely (largely) removes your own particular bubble effect, the one affecting actions on your usual devices due to your particular history of use of these devices; but it replaces it with the bubble of the other specific computer, network and community of users who worked on it before you logged on.

Now, we did spent quite a good amount of time in the previous chapters explaining that bias is a normal and altogether not too problematic feature of any ethnographic fieldwork, and that the response to it must be awareness of bias. Remember: ethnographic knowledge is inter-subjective knowledge, co-constructed by all participants in the event. Seen from that perspective, the bubble effects are mere extensions of the inevitable bias inscribed in our fieldwork practices. But let’s remove the word ‘mere’ from the previous sentence, for the extension we see is an extension in another direction – a shift, in other words. And the shift has to do with the meaning of inter-subjective in what preceded. When you interact with an online device, you’re not interacting with a particular person whose subjectivity (and, of course, bias) can be to some extent explained and understood in terms of one’s social, cultural, personal backgrounds – the ‘context’ as we know it from the literature. You’re interacting with a machine that incorporates and creates contexts that require very different modes of interpretation.

In a moment, we shall be more specific – and constructive – about this problem of online contexts. But for the moment, let’s take this on board: going online takes your field in a direction which is not in any way a direct reflection of the offline contexts you, as an ethnographic fieldworker, got accustomed to through intense interactions with the people you work with; it sends you into a different sociocultural realm and confronts you with modes of bias that are sometimes impossible to understand, let alone anticipate or predict in research. The social facts we can observe online are mediated and curated by technologies in complex synergies with their users. Overlooking this point (and it is compelling) can cause you some trouble in making sense of what goes on in the lives of the people you work with in fieldwork.

But there is more.

Who is there?

In the tradition of social research, one thing used to be quite straightforward: the identity of the people one did research with or about – the ‘population’ in one’s research. People’s identities were known, and researchers could believe that they knew them well. So well, in fact, that we could anonymize them in our research outcomes, and that we felt compelled to do so because our research had actually revealed so much about them that they could be construed as identifiable individuals. Anthropological ‘informants’ were only useful, so to speak, when a measure of intimacy had been established between the anthropologist and the ‘informant’ allowing more than mere superficial knowledge to be exchanged.

This knowledge of the population was grounded, as Michel Foucault (2008) described, in some of the great structures of modernity: nation-state bureaucracy and its elaborate inventories of people residing on the state’s territory. From birth certificates through school reports, hospital records, police files and intelligence reports, passports, tax returns, occupational, demographic and income data and the cyclical census: one of the purposes (indeed, needs) of the modern state was comprehensive knowledge of its population. An elaborate bureaucratic infrastructure served that purpose and statistics emerged as the science that could answer questions evolving from all that. As the name itself reveals, statistics was the science of the state. And statistics came up with methodologically refined tools such as the sample to turn knowledge of the population into measurable, user-friendly units with almost infinite opportunities for application.

All of this was achieved in an offline world; the present online-offline nexus offers some serious problems. The first one is infrastructural. Whereas states used to be unchallenged when it came to gathering and elaborating knowledge at a very high scale-level – that of the entire population, this monopoly has vanished. The state now competes with (and often relies upon) private corporate actors when it comes to such high-scale level knowledge. It is the likes of Google, Microsoft, Huawei, Facebook, Weibo who are the great data collectors and analysts presently: companies who collaborate with the state but who are formally independent from it, and who have the capacity to independently develop (as well as own and sell for profit) big data handling and machine learning tools and products. Knowledge of populations nowadays is distributed over more actors, many of which fall outside the raison d’état which Foucault saw as the engine behind modern population studies.

Such private actors can and do impose rules of their own – the scale level we used to define as ‘public’ is now governed by a range of different and sometimes conflicting modes of governance. And such new modes of governance deeply affect this self-evident part of social studies: knowledge about who is involved in social action.

As all of us know, the online world is populated by people operating through an alias. Trolls and members of obscure debating groups in the darker corners of the Web instantly come to mind; we also know that some online platforms are very vulnerable to interventions by automated bots and hired clickfarm operators sending out updates and responding to them; but in many cases there are also strong social and political incentives to remain anonymous when engaging in online activities. One’s employer may not be amused when an employee regularly posts social media updates criticizing the company or articulating views that can be perceived as damaging to the company’s interests; security forces may be alerted by strong political criticism voiced by people online; or one’s spouse would not appreciate one’s active presence on dating sites. In online environments where people are aware of surveillance and censorship, one’s mere presence on a forum can be experienced as risky, and participants will adjust their behavior accordingly – primarily by hiding identity features that might lead to easy identification (cf. Du 2016). The effect is: billions of online ‘profiles’ about whom interlocutors cannot assume any identity feature with any degree of certainty: the exciting 24-year old woman with whom one flirts on a dating site might actually be a 55-year old, married and quite boring man. And the revolutionary activist who eagerly invites and endorses your politically inflammatory updates might actually be a state security agent.

At the frontstage of the online world, identity uncertainty rules. The real identities of online actors are, as a rule, only known backstage by institutional actors: by internet and platform providers, the authorities and the security services. But hackers prove on a daily basis that even that level of certainty about who is online is not entirely bulletproof.

As said before, the online world provides entirely new contexts for all of us. The effects for fieldwork are momentous. While, in offline fieldwork, you can ask friends and neighbors, or colleagues and bystanders for information about particular individuals, your opportunities for doing so in online research are extremely limited – you can never be sure that the neighbor you invite to offer background information about someone is not, in effect, a neighbor at all. So as a rule, you can only observe what you see people do in online fieldwork sites. Getting feedback about who did what, however, is terribly difficult and – to add to the mess – not very reliable. For the online sources you’d approach for such feedback are almost by definition as elusive as the target of your inquiry with them. The fieldworker, consequently, is often reduced to the role of witness rather than that of investigator, and left with very few tools for upgrading one’s role from witness to investigator. So take this as a given: in online fieldwork it is immensely difficult to establish the intimate knowledge one can construct about offline respondents.

But there is more, and we need to return to the bubble effects we discussed earlier. Recall what we said there: the bubble shapes a context for social action on the basis of ‘profiles’ created by data aggregations. So here is yet another level of backstage identity construction: one not directly performed by ourselves but imposed on us by machines and influencing what we can do and effectively do online. Obviously, this also affects what a fieldworker can observe online.

Let us make this a bit clearer. The bubble brings people into your orbit whose profiles have been constructed by algorithms. These people are, also in official parlance, ‘data subjects’ constructed out of hypothetically common features based on aggregations of users’ data. As we said before, the criteria by means of which people are connected to aggregations of data are very difficult to get access to – it is safe to assume that we cannot know the grounds on which algorithms judge that certain people are similar to us, share interests, behavioral or character traits sensed to be compatible with ours, and could be brought into some kind of community alongside us. We can provide educated guesses, no more. But since bubble effects are inevitable, the upshot of all of this is that we observe very peculiar, curated social facts, full of uncertainties about who is involved in their performance. And note that the uncertainty about who is there in your online fieldwork site is individual as well as collective; it applies to the actual interlocutors whose online actions you observe, as well as to the communities that fill the bubble in which you roam.

Imagine now that you’d wish to run a survey online, using a platform such as Twitter. How will you construct a reliable sample in which sociological diacritics such as gender, age, location, education background and religion are adequately spread – when none of this can be established with certainty? How can you reach ‘everyone’ whenever you attempt to speak about a population – when you are mindful of the bubble effect? How can you even identify individual actors when the same person can have eight different Twitter accounts? And how can you be sure that ‘@EddieJones1991’ is not the 28-year old Welsh accountant he claims to be, living in Liverpool with his wife and two young kids and enthusiastically endorsing the Tories, but in fact an automated bot or a clickfarm account operated from Bangalore, India?

All of these issues about who is who online dislodge the certainties used as baseline assumptions in more than a century of social research, and they render forms of research still hanging on to such assumptions very doubtful indeed. In a moment, we shall offer some hope for ethnographically inclined researchers. But first we need to address a third major complication of the online-offline nexus.

Where are we? Invisible lines

Let us briefly recapitulate. We have seen that the online-offline nexus seriously complicates two things we used to consider rather unproblematic in offline fieldwork: what we (can) observe, and who is involved in what we observe. The bubble effect and the uncertainty about participants in social action online render both highly problematic now, and they must serve as a critical check on the kinds of claims we believe we can make in our research. There is a third obvious dimension of social action which is profoundly distorted by the online-offline nexus: the site where we perform our research.

For evident reasons, the site of fieldwork used to be perhaps its least problematic aspect. As outlined in the previous chapters, we used to choose a place for our research based on prior knowledge and a round of thorough preparatory study. Next, we would pack our gear and head off to that place. Yes, we emphasized, the actual meaning of that place would change during fieldwork as a result of accumulated knowledge – the school we chose as our site would gradually transform into a more complex habitat for those involved in the activities in that school, including the fieldworker. But in many ways, our choice of fieldwork site would define and constrict our assumptions about participants and the actions they engage in. We knew that, to stick to the example of a school, some transcontextual analysis was required, for many of the actions performed locally (and offline) by teachers, pupils and other local stakeholders would be inflected by things such as education policy, management principles and other forms of external pressure and influence. In the online-offline nexus, however, the meaning of ‘transcontextual’ has changed quite profoundly.

Two dimensions of this change need to be identified. In both instances, the guiding question is: how can we understand what goes on in our chosen fieldwork site?

The first dimension has to do with the nature of the activities we observe locally. Let us start with an anecdote. A little while ago, one of us was required to check attendances at the start of a class. The usual signup form started moving slowly through the lecture theater, and after a few minutes, suddenly two students came hurrying into the hall – alerted by their colleagues’ hastily written smartphone messages telling them that their presence was mandatory. A local action – taking attendances – was ‘exported’, so to speak, to different places elsewhere by means of online connections, and resulted in a reconfiguration of the local activity – two students joining the class.

This anecdote shows us that in the online-offline nexus, there are invisible lines connecting offline spaces with translocal ones; and that local activities are almost invariably influenced and shaped by translocal ones. Converted into the vocabulary we used above, we see how offline activities are almost invariably influenced and shaped by online ones. Such influences can be material, as in our anecdote in which a material space as well as its population get reconfigured due to online signals given by students. But even more frequent are immaterial effects of online activities on offline ones: knowledge effects, as when we cook a curry after having read several online recipes and watched some YouTube tutorials, or as when our car’s GPS system directs us to take another route due to dense traffic on the normal one. The internet is primarily a learning environment from which we extract (and on which we upload) tons of bits of information, instructions and normative judgments about how certain things should best be done (Blommaert & Varis 2015). In a formal sense also, the online world is a learning environment. Try to imagine studying without access to online resources these days, from online downloadable research publications over Wikipedia to simple Google searches – the contemporary world of learning is an online-offline one.

These learning environments have immediate effects on locally performed actions, as we have seen in the anecdote above. And these effects are inflected by the features we discussed earlier; bubble effects and algorithmically configured profiles creating peculiar forms of ‘truth’ and norms within often elusive online communities and with immediate feedback effects. To illustrate the latter: if you want to cook a Thai dish and choose, out of dozens of options, an online recipe using dried red chili rather than fresh one, this preference will be recorded in your algorithm and have an effect on your bubble. Later searches might show you more recipes using dried chili and let you interact with people who show the same preferences (unless the algorithm decides you’ve made the wrong choice and will try to rectify you in the future). In that sense, online knowledge effects may be qualitatively different from the more traditional ones. Yes, reading a book or having a conversation in a pub may have similar effects on what we think and do, but such effects were usually slower and perhaps less pervasive than the ones we currently notice in the online-offline nexus.

This is the first dimension we had to address: online resources infusing local actions and changing them due to immediate translocal involvement. The second dimension extends this somewhat and raises the question: who is involved in local actions – who belongs to the ‘personnel’ of things we observe in online life. And here, too, an anecdote can be useful as a point of departure.

Oud-Berchem is an inner-city working class and immigrant district in Antwerp, Belgium. One of the remarkable features of the neighborhood is the density of new evangelical churches, usually of the charismatic branch of protestant Christianity and run by pastors from Africa, Asia and Latin America (see Blommaert 2013). The churches are what is known as ‘storefront churches’, renting relatively cheap vacant commercial premises in an old shopping street and usually displaying a health and safety permit for 49 people. Local congregations can be slightly larger though, but some of the churches also cater for smaller congregations. Churches often change premises, denominations and constituencies – a reflex of the rapidly shifting demographics of the neighborhood.

One of the most recent arrivals in this religious industry in Oud-Berchem is a church run by a Nigerian pastor. Let us nickname the church the ‘True Religion Church of Christ’. The church rented what is probably the grottiest location in the neighborhood: a former interior decoration shop closed down a handful of years back, quite badly affected by years of vacancy and exposure to the elements. The church has a permit for 49 attendants, and this is about the size of the congregation attending Saturday and Sunday services there. It’s a small, hardly remarkable and even less prestigious enterprise.

Our initial research on the neighborhood and its churches was based on traditional – read: offline – ethnographic linguistic landscape analysis. From that perspective, indeed, the True Religion Church of Christ is a small local phenomenon, eclipsed by other churches with more attractive premises and a larger congregation. At a given moment, however, we started paying more attention to an often overlooked feature of the linguistic landscape: website addresses and social media signs of the ‘Follow us on Facebook’ type (Blommaert & Maly 2019). When we followed such pointers for the True Religion Church of Christ, we bumped into a few surprises. Its pastor turns out to be a modest global celebrity in the domain of charismatic protestant religion. He runs a YouTube channel with over 125,000 subscribers; the main feature video there is one in which the pastor brings a dead boy back to life during a service in Nigeria, attended by many hundreds of faithful. This video was watched over 85,000 times. The pastor also runs a website in which he announces services all over the world – North America, Europe and Africa – and through which items and services can be booked and ordered using standard e-payment methods such as Paypal.

Suddenly, the grotty premises in which the local congregation gathers on Saturdays and Sundays appear in a different light: as a mere node in a global network of religious activities connected by advanced online infrastructures. This global network is big and prestigious, and stands in sharp contrast to the smallness and shabbiness of what goes on in Oud-Berchem. Many more people, places and resources are involved in what goes on in Oud-Berchem than those that can be locally observed. And we can reasonably assume that what goes on locally in the True Religion Church of Christ derives a lot of its meaning and impact from the translocal, prestigious and well-resourced network in which it is one local node and to which it is permanently connected by online infrastructures. In fact, what happens locally is probably possible only because of the existence of this larger network and its online resources. And so, when we observe the local activities of the church’s congregation, we need to be aware of the fact we see just a very small part of the total social fact we need to understand, and which we can engage with by following the pointers that take us online.

So here are the two dimensions we needed to bring up: the fact that offline practices are almost invariably influenced, formatted and enabled by online ones; and the fact that locally performed social actions can involve far more people than those actually present locally – the effective personnel of lots of current social actions can only be gauged by connecting the offline local phenomena with the online translocal ones. Our field has effectively become an online-offline field; doing fieldwork requires presence in and attention to both, and the blissful simplicity of ‘the local’ has been traded for a far more complex reality of connected fieldwork sites. The notion of ‘participant observation’ needs to be literally in the online-offline nexus: ethnographers are participating in exactly the same contextualized processes they are studying, and there is no privileged vantage point that gives us and edge over other, ‘ordinary’ participants.

More complexity? More ethnography please!

All of this is bad news of course. In the online-offline nexus, we are forced to surrender some of the things we long thought were relatively simply: the things our field had to offer in the way of observable facts and information, the people with whom we engaged in fieldwork, and the actual sites of fieldwork. In other words: we need to reconsider the what, how and where of fieldwork. The online-offline nexus, we can see, is quite a bit more complex than the good old traditional offline fieldwork arena.

The bad news, however, is mainly for those branches of science that rely heavily on the assumptions we questioned above. And there are several reasons why ethnography, while needing to be cautious and more than just aware of these changes, is best equipped to deal with them. In chapter 2, we explained that ethnography is a scholarly approach which, in contrast to many other approaches, does not attempt to simplify and reduce complexity; it takes complexity as a point of departure and tries to provide a full and detailed account of it. Ethnography is not about removing the chaotic nature of social practices performed in real, concrete contexts – it is about making sense of that chaos. The fact that the chaos appears to become denser in the online-offline nexus should not deter us: it’s still just chaos, and we must make sense of it.

And we have inroads into it. Even if the what, who and where of fieldwork are getting more complicated, there are things we can reliably observe. We can still observe what people do, the social actions they perform. In fact – and we emphasized that as well in the opening chapters of the book – ethnography is focused on making sense of social action, of concrete social action performed in concrete contexts, and it belongs to that broad tradition in social research captured under the umbrella of the ‘action perspective’ (cf. Blumer 1969; Goodwin & Goodwin 1992; Strauss 1993; Rawls 2002). So we can observe people watching online stuff, doing online searches, asking and responding to questions, telling stories, making an argument, insulting or responding to insults, expressing joy, appreciation and gratefulness, grief, anger, uneasiness, concern, irony and humor, thanking others; we can observe them liking, sharing and reposting, commenting and endorsing or distancing themselves; we can observe them incorporating online material produced by others in their own online interventions; we can see them logging on and logging off; subscribing to channels and profiles and blocking or ignoring others. And we can observe the (largely visual, literate) resources they deploy in doing all that: different forms of language, jargon and slang, different forms of writing, emojis, memes, GIFs, selfies, profile and banner images, video chats and livestreams on a variety of apps – name it. All of this, we know, is done in interaction with others, frontstage as well as backstage – one is never alone on the Web – and mediated by the specifics of the online contexts we laid out above.

That’s a lot. In fact, it’s exactly the stuff needed for ethnographic work, as we explained in chapter 2 of the book. And it is by looking at the intricate interplay between actions and resources that we are able – in ethnographic analysis – to see how people navigate the contextual opacity and the identity uncertainties that characterize online interactions and make sense of that chaotic reality (cf. Szabla & Blommaert 2017), how they engage in the learning processes for which the online world offers such infinite opportunities, and construct identities and communities within their bubbles, and often beyond them (Varis & Blommaert 2015; Prochazka & Blommaert 2019).

So it is not because we cannot observe everything in online contexts that we can observe nothing. We cannot observe the algorithms and surveillance systems that create bubbles and profiles, true. But we can observe the ways in which people engage with them and operate within their confines – how they adjust their social conduct to the complex and largely invisible contexts within which they interact with others. This is an eminently adequate ethnographic object of inquiry.

But we need to address it carefully. Whenever the phrase ‘participant observation’ was used in discussions of fieldwork, the focus used to be on ‘observation’, and it carried the suggestion that, while participating in social processes, ethnographers did something special and did that from a privileged position – they ‘observed’. We believe that fieldwork in the online-offline nexus shifts that focus towards ‘participant’, and that we must forget the possibility of a privileged position of observation. Whatever we observe is observed as a participant in a new field in which breaking out of the contexts of ordinary participation is near-impossible, for important aspects of such contexts are impossible to inspect – the backstage aspects we discussed above. Perhaps this was never possible, even in traditional offline fieldwork, and perhaps it was just (in Johannes Fabian’s (1983) famous view) the conventional arrogance of academia that created the claim towards privileged knowledge positions. In that case, the online-offline nexus confronts us with an unpleasant truth – one which renders our work more complex but equally more interesting.



Blommaert, Jan (2013) Ethnography, Superdiversity and Linguistic Landscapes: Chronicles of Complexity. Bristol: Multilingual Matters.

Blommaert, Jan (2018) Durkheim and the Internet: On Sociolinguistics and the Sociological Imagination. London: Bloomsbury.

Blommaert, Jan & Ico Maly (2019) Invisible lines in the online-offline linguistic landscape. Tilburg Papers in Culture Studies, paper 223.

Blommaert, Jan & Piia Varis (2015) Enoughness, accent and light communities: Essays on contemporary identities. Tilburg Papers in Culture Studies paper 139.

Blumer, Herbert (1969) Symbolic Interactionism: Perspective and Method. Berkeley: University of California Press

Du Caixia (2016) The Birth of Social Class Online: The Chinese Precariat on the Internet. PhD diss., Tilburg University.

Fabian, Johannes (1983) Time and the Other: How Anthropology Makes its Object. New York: Columbia University Press.

Foucault, Michel (2008) Security, Territory, Population. Lectures at the College de France 1977-1978. London: Palgrave Macmillan

Goodwin, Charles & Marjorie Harness Goodwin (1992) Context, activity and participation. In Peter Auer & Aldo DiLuzio (eds.) The Contextualization of Language: 77-99. Amsterdam: John Benjamins.

Kaur-Gill, Satveer & Mohan Dutta (2017) Digital ethnography. In Christine Davis & Robert Potter (eds.) The International Encyclopedia of Communication Research Methods: 1-11. New York: Wiley.

Prochazka, Ondrej & Jan Blommaert (2019) Ergoic framing in New Right online groups: Q, the MAGA kid, and the Deep State Theory. Tilburg Papers in Culture Studies paper 224.

Rawls, Anne Warfield (1987) The Interaction order sui generis: Goffman’s contribution to social theory. Sociological Theory 5/2: 136-149.

Rawls, Anne Warfield (2002) Editor’s introduction. In Harold Garfinkel, Ethnomethodology’s Program: Working Out Durkheim’s Aphorism (ed. Anne Warfield Rawls): 1-64. Lanham: Rowman & Littlefield.

Strauss, Anselm (1993) Continual Permutations of Action. New Brunswick: Aldine Transactions

Szabla, Malgorzata & Jan Blommaert (2017) Does context really collapse in social media interaction? Tilburg Papers in Culture Studies paper 201.

Varis, Piia & Jan Blommaert (2015) Conviviality and collectives on social media: Virality, memes, and new social structures. Multilingual Margins 2: 31-45.

Varis, Piia & Mingyi Hou (2019) Digital approaches in linguistic ethnography. In Karin Tusting (ed.) The Routledge Handbook of Linguistic Ethnography. Abingdon: Routledge (in press).