Nearly 30 years ago, empirically sound corpus-based research on language variation in the media seemed an elusive goal. In a Special Issue of the Annual Review of Applied Linguistics, one of the founders of sociolinguistic research on the media, Allen Bell, painted a rather pessimistic picture of the entire enterprise, deploring the paucity of analytical tools and the time-consuming methodologies that targeted “full-scale analyses of only a few texts, alongside more piecemeal, specific analyses on larger samples” (Bell Reference Bell1995:33). He deemed the complexity of the theoretical frameworks of his time “daunting to students or lay people (and sometimes to specialists)” and sociolinguistic research on the media in “danger of basing conclusions on inadequate language analysis” (idem). He, nonetheless, found that the study of language and discourse of the media had already “gained a coherence and focus it previously lacked” (24), and its importance was uncontestable “for both what it reveals about a society and for what it contributes to the character of society” (23). Echoing scholars in communication, he acknowledged that while sociolinguistic studies of media discourse were still rare at that time, when “linguists set their minds and skills to the task, the dividends [could] be rewarding” (34).

Fast forward to the present when widespread digital communication offers unprecedented opportunities for synchronous interaction to hundreds of millions of people worldwide, and it is probably fair to say today’s sociolinguists are much better equipped to take up the challenges singled out by Bell. Thanks to new analytical frameworks and a rich array of computational tools, some old and new research questions can finally be addressed. Could the large number of spoken forms and the shortening of words and expressions in chat and text messages be due to “semiotic compensation” for the lack of verbal and visual input (Androutsopoulos Reference Androutsopoulos, Coupland and Kristiansen2011:149)? Are new terms and orthographic forms called ‘net neologism’ and ‘neography’ in the messages of young SMS users – that “tend to be social in purpose and informal in style” (Anis Reference Anis, Danet and Herring2007:88) – true lexical innovations or old spelling games in new guises? Given the spectacular increase of informal writing in the digisphere, is it time to celebrate the emergence of “internet linguistics” as a new subfield of linguistics (Crystal Reference Crystal2011)?

Papers in this Special Issue, while connected to these overarching questions, represent a well-organized collection of empirically grounded, corpus-based approaches to language variation in interactive digital media. They fill a gap by providing a coherent and yet diverse sample of studies on a variety of structural aspects of a single language: French. The thematic focus of the volume reflects a traditional framing of language variation, well known from variationist sociolinguistics, correlating empirically observed phonological, morphological, and lexical patterns with social and linguistic variables, among them age, gender, and genre. In this commentary, I will reflect briefly on each paper’s unique contribution to the field as well as the volume’s main goal to examine socially conditioned variation in selected contexts of twenty-first-century digital writing.

“The [existence of the] written media – including the new social media – and, for the last century or so, the electronic spoken media, means [that] lexis can be spread at a distance also,” Trudgill (Reference Trudgill2014:215) reminds us, with respect to the role of the media in disseminating new words. And yet, few studies to date have shown how large-scale repertoires and community norms are affected, if they are at all, as these long-range processes have been “very difficult to capture empirically” (Androutsopoulos Reference Androutsopoulos, Cotter and Perrin2017:417). In principle, media influence doesn’t necessarily need to lead to population-wide, or even systemic, changes, as it doesn’t always require regular interaction between speakers and can involve ready-made lexical material “taken right off the shelf” (Eckert Reference Eckert2003:395).

As it turns out, two papers in this volume bring new insights to this very question. Blondeau and Tremblay’s work on multiple sociolinguistic variables in text messages in Quebec provides evidence of a new vernacular genre of writing that is both coherent and independent from written Standard French and allows for the ‘enregisterment’ (Agha Reference Agha2007) of innovations, including orthographic and lexical forms, with new semiotic meanings. Zsombok’s investigation of the uptake and diffusion of internet terminology through the competing use of mot-dièse and mot-clic, two French lexical variants of the English word ‘hashtag’, adds to this perspective by bringing into close proximity empirical data and linguistic ideologies. His corpus allows the tracking of purist attitudes, and the lack of thereof, towards these variants, revealing their differential treatments by Quebecois and French Twitter users. Echoing recent findings on digital communication in diasporas of polycentric languages (Androutsopoulos and Lexander Reference Androutsopoulos and Lexaner2021), these papers suggest that each of these centres of Francophonie possesses a unique profile of linguistic and semiotic resources conditioned by different constraints and rooted in different sociolinguistic hierarchies.

The remaining four papers in the volume examine the extent to which digital writing, especially ‘neography’, is harvested as a semiotic device to convey social meaning associated with variable phonological phenomena: dialectal variation in the realizations of phrase-final consonants and vowels (Lamontagne and McCulloch), dialect-specific nasal vowel allophony (Law), the use of diacritics as a function of age (Williams and Blattner), and the relationship of phrase-final fricative epithesis (PFFE) to gender, style, and affect.

Lamontagne and McCulloch’s results on ‘letter multiplications’ shed new light on the simultaneous expression of affect and systemic phonological variation by way of neography. They reveal, among other original findings, that speech prosodic information is computed by Twitter users when they type letter sequences repeatedly in a French Tweet to convey emphasis (e.g. <ouiiiii> for oui ‘yes’). Furthermore, dialect-specific patterns in the realization of word-final vowels (realized in Midi but not in Northern Metropolitan French) also come across in patterns of letter repetitions, showing that Twitter users continue to ‘think like speakers,’ i.e., having actual speech segments in mind, when they write. The authors’ final discussion reinforces the impression of an ‘infusion of orality’ into the written digital vernacular, as they speculate on the role of sonority in the lengthening of word-final consonants and the precise acoustic correlates associated with letter repetitions in their corpus.

Law’s paper continues a long tradition in French sociolinguistics of tracking representation of ongoing phonological changes in innovative spelling (e.g., Malderez Reference Malderez2000 by school children, van Compernolle Reference van Compernolle2008 in online chat). Its results point to parallel findings in speech perception research, but they also reveal that structural variation on Twitter is impacted by factors unique to the written modality. Two important differences uncovered in the paper indicate that lexical frequency is a major conditioning factor in the use of innovative spelling and, ongoing mergers and shifts in nasal vowel contrasts are well above the level of awareness in some cases, as they are perceived, spelled, and readily discussed by some Twitter users.

Williams and Blattner’s replication study is a heavily quantitative take on the use of diacritics by multiple generations of online chat users in French. The authors confirm that the use of the graphemes é, à, and ç remained stable between 2008 and 2016, but they also show that communicative practices had shifted over time, and not necessarily in the expected direction: in 2008 diacritics were favoured by ‘the fifties (cinquantaine)’ age group, but by 2016 ‘the twenties (vingtaine)’ group was the most comfortable typing the largest number of letters with accent marks. Acknowledging that these apparent-time patterns remain difficult to interpret, especially in the absence of verifiable biographic data from the participants, they advocate for a qualitative dimension to corpus-based statistical analyses of digital texts.

Based on a Twitter corpus of a 12-year span, Dalola’s paper documents hitherto unknown aspects of phrase-final fricative epithesis (PFFE), a sociophonetic variable in Hexagonal French that consists of adding fricative-like sounds (here, letter sequences) to words that end with high vowels (e.g., beaucoup_h, oui_ch). Two of those are: intense affect and social stereotypes. The results lead us to believe that, curiously enough for phrase-final fricative epithesis, phrase-finality is not among the strongest conditioning factors of PFFE in digital communication on Twitter. One of the likely reasons could be differences between written and spoken modalities: due to the low character count, Twitter messages can never display the full prosodic hierarchy and, thus, the number of variable prosodic contexts in which PFFE can occur is reduced and cannot condition variable use. Prompting us to refine the interpretation of ‘orality’, one of the “three main themes of innovation and change” in digital communication (Androutsopoulos Reference Androutsopoulos, Coupland and Kristiansen2011: 154), the paper shows that Twitter might seem like orality ‘poured into a written mold’, but in terms of structural patterns of variation, it appears to be an independent modality in its own right.

Collectively, the six papers of this volume underscore important advances made in sociologically oriented, structural analyses of vernacular writing in the digital era. They provide sound theoretical frameworks and replicable models of large-scale, corpus-oriented studies; ones that Bell and his pioneering generation of sociolinguists studying language in the media might not even have envisioned several decades ago. The papers also leave us with some unexpected findings that point to contradictions and inspire new hypotheses. One of them is that, contrary to expectations, dichotomic distinctions of scale (see Blommaert Reference Blommaert2010), whereby ‘old’ vernacular writings would be locally bound genres and ‘new’ vernacular writings would escape strong ties to local speech patterns, are very likely an oversimplification. Several papers in this volume attest to the emergence of systemic variation that seems to have created new structures and genres proper to the digital written medium itself. Thus, as “more people write, people write more, and unregimented writing goes public” (Androutsopoulos Reference Androutsopoulos, Coupland and Kristiansen2011:154), the elaboration of digital vernaculars does not simply “turn writing into a medium of sociability” (idem), but rather leads it down on a path of a new communicative genre.