1.3.1 Corpus-Assisted Discourse Studies (CADS)

Corpus-assisted discourse studies is a methodological framework that moves the focus from the technicalities of the corpus investigation to the exploration of language in its relation to aspects of society and culture. Specifically, CADS ‘seeks to capture the recurring traces left by social routines [where] the starting point is not linguistic but social’ (Taylor and Marchi, Reference Taylor and Marchi2018: 61).Footnote ⁸ That is, researchers could be motivated by issues of social justice (as reported by Baker Reference Baker, Taylor and Marchi2018, in what he refers to as action research). While the combination of corpus linguistics and discourse is not new, the name CADS allows scholars to frame their work and be part of a research community. I do not mean to say that it is a rigid framework with a restricted community; on the contrary, it is a dynamic tool through which research can be conceptualised at the intersections between corpus linguistics and discourse. I am assuming that the readers are familiar with the debate on the notion of discourse; here, I borrow the explanation from the most recent publication on CADS (Gillings, Mautner, and Baker, Reference Gillings, Mautner and Baker2023: 1): language is ‘analysed as performing social functions’. Acarno (Reference Acarno, Fina and Georgakopoulou2020) also notes that CADS focuses on discourse studies rather than discourse analysis, with the scope of conveying the interdisciplinary nature of the framework. Here, I wish to bring some aspects to the fore: (i) the inclination to use CADS to explore social issues, (ii) its flexibility about where one positions their research on the two sides (Corpus Linguistics and Discourse Studies), and (iii) the implications of this concerning power and ideology/ies. I concur with what Taylor and Marchi suggest (2018: 61) in that CADS is interested in uncovering insights into a ‘particular situation, purpose or function repeatedly enacted within a speech community’. The relation between society and language is seen in analytical explorations from text to discourse (Partington et al., Reference Partington, Duguid and Taylor2013), in strict relation to what occurs in the society investigated through language phenomena, as well as in the simultaneous exploration at micro and macro levels (Acarno Reference Acarno, Fina and Georgakopoulou2020). To do so, the quantitative aspect is not left without several in-depth qualitative remarks, which is why I choose this framework for analysing gender-inclusive language. The term ‘quantitative’ is perhaps a simplistic way to discuss some philosophical underpinnings such as recurring patterns, regularity, and, to some extent, what Partington et al. (Reference Partington, Duguid and Taylor2013: 10) refer to as serendipity. Recurring patterns tell us that what is happening might be established in the corpus as a representation of what is outside the corpus. Similarly, regularity can foreground what happens, putting the data set at the centre rather than the analyst who is called to interpret it rigorously. Differently, serendipity can ‘show us things we perhaps didn’t even know we didn’t know’ (Partington et al., Reference Partington, Duguid and Taylor2013: 9), allowing the analyst to share insights with the scientific community and communities. For these reasons, building a CADS corpus means foregrounding the research’s society-related scope while also maintaining the principles of representativeness – that is, selecting texts that are as relevant as possible to the phenomenon or the variety chosen. Partington et al. (Reference Partington, Duguid and Taylor2013) argue that one difference with corpus linguistics lies in CADS analysts exploring external data as a source of information to interpret but also to analyse the corpus. The corpus is not a black box (Partington et al., Reference Partington, Duguid and Taylor2013: 12) but a source. The way in which this source is used to extract and interpret discourse is based on the flexibility of this framework, where research procedures, research goals, and, to some extent, levels of subjectivity are bespoke to each investigation. With regard to the opportunities and the affordances CADS offers, one can argue that it ‘requires as much commitment to the computer-assisted profiling of corpora as to the human-led investigation of those discursive phenomena which are beyond the reach of automated analysis’ (Gillings et al., Reference Gillings, Mautner and Baker2023: 7), where the analysts decide the ratio between the two based on the research question and the scope of the research (Acarno, Reference Acarno, Fina and Georgakopoulou2020). I argue that the analysis of gender-inclusive language in this Element expands some of the assets of CADS – for instance, the reproduction of habitual patterns (Taylor and Marchi, Reference Taylor and Marchi2018). My corpus is interested in how society deals with, embraces, and explores a new language phenomenon within a specific social (media) context. Furthermore, I conceive of the corpus as an ethnographic site (see Section 2), expanding on the notion of the relevance of the corpus in CADS investigations. In this brief overview, I attempted to present the main elements of this framework, and developments from these major points will be made across the Element. However, one aspect still deserves attention: in CADS tradition, discourse is not conceived of through a specific lens – for example, Historical Discourse Analysis or Critical Discourse Analysis. This should, therefore, not be seen as a shortcoming but rather an opportunity, as the one I wish to grasp here, to explore philosophical, methodological, and theoretical connections with established or new frameworks. For this reason, I considered studying the synergy with FCDA.

1.3.2 Feminist Critical Discourse Analysis (FCDA)

Feminist Critical Discourse Analysis has been theorised by Michelle Lazar (Reference Lazar and Lazar2005, Reference Lazar2007, Reference Lazar, Ehrlich, Meyerhoff and Holmes2014) and has gained some attention yet, I argue, it remains unexplored in language and gender (with some exceptions; see Nartey, Reference Nartey2024). I say gender because Lazar’s work focuses on women and men as a starting point to explain how this framework can provide a platform for discussing contexts and language investigations. However, in her work, one can find room to expand, mould, or widen, if not the scope, the operationalisation of research. Lazar is firm in arguing that FCDA is not simply gender and Critical Discourse Analysis together, even though it borrows the focus on the complex workings of both power and ideology; FCDA foundations are patriarchal imbalances, gendered assumptions, and gender-related power asymmetries. In other words, these are not detached from the theoretical framework as they are embedded in it. In reworking feminist theoretical underpinnings, Lazar (Reference Lazar, Ehrlich, Meyerhoff and Holmes2014: 182) explains, FCDA is deep-rooted in a ‘feminist political imagination’, allowing how society operates (mainly from a discriminatory point of view) to emerge in language investigations. Some principles are at the core of FCDA, which I see aligned with the research proposed here. The first principle, feminist analytical activism, is about conducting research not only to talk about the material consequences of the imbalances to take but to see the published work ‘as action’ (185). The second principle, gender as ideological structure and practice, is based on the notion that society is divided into gendered groups, not based on biology but on values associated with men and women and, in my work, on a broader understanding of gender and sexuality that moves away from the binary. These two principles are fundamental to understanding how this Element relates to the reality outside academia and research, as well as my positioning as a feminist scholar. The third one, the complexity of gender and power relations, is central to what I am exploring in this Element. While in the first two, Lazar emphasises discourses of covert or overt discrimination (e.g., sexism) to explore in the language used in a specific cultural context, the third one opens to discourses of resistance and challenge to the status quo, which, in my view, is what gender-inclusive language sets out to do. Lazar recognised that this side of the framework had not paid equal attention to discourses of discrimination, making it an opportunity for the research conducted here to expand on the usefulness of FCDA. Resistance and challenge discourses must be seen as empowering and situated in a specific, large or small site of struggle. The fourth principle, discourse in the (de)construction of gender, builds a bridge between this framework with theoretical underpinnings such as poststructuralism (briefly put, language is a resource which is socially shaped); in addition to this, Lazar mentions the relevance and salience of allyship, yet not using this term. In the discussion of this principle, mentions of methods are given; as in the tradition of other critical studies, FCDA does not recommend a specific method or methodology, as it is concerned more closely with how the findings are read in the analysis of the social, cultural, and political context. The last principle, critical reflexivity as praxis, is linked to the first one in that it asks the researcher to transfer the knowledge in a variety of modus operandi and several contexts, providing a ‘rich and powerful critique for action’ (Lazar, Reference Lazar2007: 144). Considering these five principles, one can see how FCDA puts gender at the centre of research, allowing researchers to continue the tradition beyond known gender and sexuality binaries. In Section 1.3.3, I briefly discuss what triangulation is and how I move to achieve so based on the two frameworks, CADS and FCDA.

1.3.3 Triangulating CADS AND FCDA

This Element proposes a novel synergy between the two frameworks outlined previously, aiming to use their strengths to explain the findings. While the term ‘synergy’ gives the idea of connection, studies in linguistics (and other fields) use the term ‘triangulation’. This is not new, as some underpinnings explained here will suggest, but notable differences can be found across disciplines. Egbert and Baker (Reference Egbert and Baker2020) state that triangulation ‘is a research approach that takes two or more perspectives’. Bazeley and Kemp (Reference Bazeley and Kemp2012: 55) argue that the approaches must have ‘a common purpose that goes beyond that which could be achieved with either method’. This is paramount in this research: by reading the findings through these two frameworks combined, I wish to offer a close account of what happens when gender inclusive language is used. Flick (Reference Flick2017) convincingly pinpoints one core value, borrowing from Denzin (Reference Denzin1970[1978]) – that is, triangulating is a strategy of validation, and its use carries a maximisation of validities in the fields of inquiry concerned. Validating is intended here as a seal of consistency in reviewing the findings according to theoretical and methodological tools. While many (in corpus linguistics studies, see Egbert and Baker Reference Egbert and Baker2020) have worked towards a taxonomy of triangulations – for example, methodological, theoretical, and investigator – I am here faced with a mixed triangulation; FCDA is mainly theoretical, while CADS focuses on fruitful corpus techniques in investigating social issues. It is, in my view, the reason why these two triangulate robustly: this research did not wish to be an investigation of language, gender, and sexuality through corpus tools (and its techniques) but instead to have gender (as a broad term) as a core justification for the choice of topic, data, methods, and analysis. In other words, FCDA complements the examination of social issues (as for CADS focus) by providing a rationale for the researcher and for what emerges in the imbalanced site of struggle in which language, but primarily, gender (as an umbrella term) operates. The traces to investigate through CADS in a particular situation are for FCDA the structures in which there are asymmetries, assumptions, and resistance related to gender. In other words, the linguistic performance of social action in CADS is found in these deep-rooted workings of gender in society. They cannot be seen as separate from a historical legacy.

Furthermore, in both frameworks, the researcher has an active role in engaging (through the analysis and for FCDA in praxis) with the social topic from a justice point of view. The micro and macro levels (as for CADS investigations) are related to the fourth FCDA principle, that of deconstruction. In addition to this, CADS allows the analyst to choose the inclination to more corpus-relevant findings or discourse findings. FCDA is employed here to resolve the balance between the two (as seen in the analysis of functions, Section 2.5.3). With FCDA being theory-focused (not recommending any methodological framework) and CADS being more method-focused (not recommending any theoretical framework), I believe that some interesting parameters attributed to methodological triangulations might be challenging to apply – for example, convergent or correlational, independent, sequential or cyclical (for an overview see Egbert and Baker Reference Egbert and Baker2020). The triangulation here is more closely related to contributions in social science studies in that the goal is to ‘add breadth or depth to our analysis’ (Fielding and Fielding, Reference Fielding and Fielding1986: 33) and, as Flick (Reference Flick2017) suggests, to understand how the perspectives, placed side by side, are valuable and influential in the production of extra knowledge.Footnote ⁹ I will operationalise the underpinnings of FCDA and CADS in Section 3, providing corpus-assisted analysis of patterns, metalanguaging, and functions related to using the gender-inclusive strategy of the schwa in Section 2.

2.4.1 The Schwa-Corpus

To collect the corpus (and the sub-corpora), I decided against compiling or using a general corpus, instead focusing on a specific corpus of tweets which contained the schwa.Footnote ²³ In CADS (and CL) jargon, this is referred to as a specialised corpus (Acarno, Reference Acarno, Fina and Georgakopoulou2020; Gillings et al., Reference Gillings, Mautner and Baker2023). In this specialised corpus, I see the tweets as ‘texts’ (Egbert and Schnur, Reference Egbert, Schnur, Taylor and Marchi2018: 159) possessing three criteria: (i) naturally occurring – that is, genuine speech; (ii) recognisably self-contained – that is, language is cohesive as a unit and proper to the context/register in which it appears; and (iii) working functionally, where texts ‘do not occur randomly or haphazardly. Rather they can be characterised by communicative functions that are intended by the author/speaker and interpretable by the reader/listener’ (Egbert and Schnur, Reference Egbert, Schnur, Taylor and Marchi2018: 162). This last point is paramount to collecting my data set, allowing the triangulation between CADS and FCDA to blossom. In order to have this specialised corpus, I decided to list some words which used the schwa on Twitter starting from observing four years (from 2019 to August 2022), using the advanced search on the social media and selecting fields ‘word’ (ə), ‘language’ (Italian) and ‘dates’ (2019, 2020, 2021, 2022). I compiled the list of words with two PhD students at the University of Brighton, Ashley Reilly Thornton and Chara Vlachaki, who were hired through the university’s postgraduate research scheme. The criterion was to list all words appearing with the schwa (short and long) divided into years. Once the list was completed, I excluded adjectives (e.g., rispettosə/respectful, inclusivə/inclusive, inglesə/English, oncologicə/oncological), past participles (e.g., andatə/gone, invitatə/invited), and some nouns (e.g., pazientə/patient/s), in order to have a consistent corpus. I also noted that the long schwa was rarely used to indicate the plural, and the short schwa functioned for both singular and plural forms.

Furthermore, some speakers seemed to have used, yet not consistently, other orthographic signs to indicate the schwa, such as 3 (e.g., inclusiv3), perhaps due to the initial obstacle of not having the ə easily accessible on the Italian computer keyboards or that of the phone.Footnote ²⁴ For this reason, my analysis does not include the use of 3 as an inclusive device. To obtain the corpus of the twenty-five schwa-words selected, I used anaconda prompt (iniconda3) in August 2022. Through a command, I requested the list of tweets, including the function (snscrape), the file format (jsonl), and the site where to scrape the words (twitter-search), followed by the terms in the list (Table 4). One crucial point is that I did not restrict the period under investigation, meaning that the scraper could capture all occurrences. The program created text files which contained each tweet and other information (what in corpus linguistics are usually called ‘metadata’). I present an example in Figure 1, where some metadata elements have been replaced so that the tweet(er) could not be identified – for example, url.

Figure 1 Example of Tweets and metadata provided by Anaconda

In jsonl format, these files contain helpful information to locate the tweet (e.g., the date, the device, the number of followers of the person who tweeted, replies, quotes, etc.). I then used a downloadable program, OpenRefine, to transform the jsonl files into Microsoft Excel files.Footnote ²⁵ Open Refine allows for the extraction of a vast range of metadata among those available from the original files. For this study, I selected four elements, which descriptions (in Table 2) are borrowed from Di Cristofaro (Reference Di Cristofaro2023: 192–8).

This metadata information, as well as the tweet (in the form of renderedContent), was useful to have some ideas about when the terms were used (date) and if specific hashtags could be useful in understanding the data set (hashtags.*). The files underwent cleaning, intending to investigate only genuine tweets written by the speakers rigorously, and the following were removed:

• Duplicates of tweets – for example, when speakers had sent the same tweet to different people.

• Tweets where the schwa-word was used in a language other than Italian – for example, carə (meaning way/option in Azerbaijani).

• Tweets where the schwa-word was the title of a newspaper/magazine article on the topic of inclusive language.

• Tweets where the schwa-word was not used as an inclusive strategy.

The Excel files produced were employed to familiarise with the data set, as in the CADS tradition, showing that no hashtag was consistently used. The list of terms selected (see Table 3) urged for some thinking to make the analysis viable and consistent, resulting in a categorisation of mainly semantic fields (Activism, Kinship and friendship, School, Work), but also one grammar-oriented field (Pronouns and adjectives).

This categorisation had to consider some semantic insights – for instance, the word compagnə has different meanings, and therefore its occurrences were split across the relevant categories: partner (in Kinship and friendship), comrades (in Activism), and schoolmates/university mates (in School).Footnote ²⁶ Initially, the categories School and Work were put together; however, in familiarising myself with the sub-corpora, I noticed that the age (of the speakers) might be a factor worth considering. Specifically, younger users probably employ terms that refer to school or university more often than professional terms (e.g., colleghə). While an apparent time approach – that is, selecting different age groups in the present to investigate linguistic change across years/periods and in the past (Bell, Reference Bell2006) – is not the focus of this research, this could provide some interesting insights. In the category Kinship and friendship, ragazzə and ragazzinə have the same meaning: young people; however, the latter contains the suffix –inə, which usually adds the underlying sense of innocence (in positive and negative prosodies). In relation to what was suggested in Section 2.2 about the formation of the words, the terms autorə (author/s), scrittorə (writer/s), and attorə (actor/s) appear on Twitter regardless of the criticisms concerning the masculine suffix (tor–). These are included in the category Work, which is the biggest in the number of terms (both in my corpus and in the words observed in the initial list). In Table 4, I report the number of occurrences for each word in the ad hoc Excel file, as part of the categories outlined previously in this Element, and the date on which the first tweet appeared in the schwa-form.

Note that this is not the final number of words for each of these terms, as occurrences of one term could also be found in other Excel files or used in association with other terms – for example, mieə amicə (my friends). Furthermore, the scraper found 32,836 occurrences of tuttə (Eng. all/everybody, 6 occurrences in 2018, 10 in 2019, 440 in 2020, 12,030 in 2021, and 1,999 in the months of 2022 that I am investigating). Starting from this, I decided to extract a sample of 5,000 occurrences (2,000 from 2021 and 3,000 from 2022) based on an average of the other two high-frequency terms, amicə (5,491) and ragazzə (4,667). This corpus must be considered a sample in its methodological and theoretical implications. On this topic, Egbert and Schnur (Reference Egbert, Schnur, Taylor and Marchi2018) encourage researchers to establish an appropriate sampling unit based on the aims of the research but also to allow a solid investigation. Limitations and pitfalls are inevitable and are noted where they occur.

Similarly, it is undeniable that cherry-picking has occurred at different levels, as described by Baker (Reference Baker, Taylor and Marchi2018: 282), but attempts were made to build and examine the corpus consistently. In addition, some observations can be made – for instance, forms such as those included in the categories Pronouns and adjectives (6,786, only including a sample of tuttə) and Kinship and friendship (11,801). These are the ones mostly rendered inclusive by Twitter users. These terms seem primarily used in informal and everyday language, descriptive of personal experiences. As in the CADS tradition, the corpus investigation is not solely based on frequency but rather on patterns originating from the corpus selected and with (possibly meaningful) low-frequency occurrences in sight (as in the serendipity explained in Section 1). In observing the dates on which these schwa-forms were first used, one can notice that most of them appeared on Twitter in the second part of 2020, possibly following the social media debate on the schwa. Only three terms were found to be used earlier – figliə (2018), ragazzə (2019), and tuttə (2018) – suggesting that some speakers were already renegotiating traditional and binary gendered morphemes. In Table 5, I present the number of tokens of the final corpus used for the investigation, referred to as the schwa-corpus, as well as the information about the sub-corpora referring to the five categories discussed earlier.

The corpora are different in size, and for this reason, statistical measures will be used in investigating frequency, collocations, and bigrams. The corpus tool used to investigate the corpora is Lancsbox 6.0 (Brezina, Weill-Tessier, and McEnery, Reference Brezina, Weill–Tessier and McEnery2020). Lancsbox is a free corpus tool for exploring languages, among which is Italian, also providing a tree tagger for parts of speech (POS). Before conducting the analysis, I checked how the tree tagger reacted to the schwa, as the assumption is that the tagger has been instructed to recognise the traditional binary morphemes. With the aim (or rather, hope) to investigate colligations – that is, ‘[a] form of collocation which involves relationships at the grammatical rather than the lexical level’ (Baker, Hardie, and McEnery, Reference Baker, Hardie and McEnery2006: 36), I conducted some queries to see how the words with the schwa were categorised. Possibly unsurprisingly, I noticed that there was no consistency in how these were tagged, as the following examples show:

• Essə was tagged as a noun and not as a pronoun.

• Solə was tagged as a noun and not as an adjective.

• Colleghə was tagged as an adjective and not as a noun.

• Lə was tagged as a noun and not as a determiner.

These examples raise methodological concerns as well as technological ones. I hope this Element and those following can initiate a discussion about how a tree tagger can account for inclusive forms, providing the same technological affordances used by those investigating noninclusive language.

2.4.2 The Corpus as an Ethnographic Site

In building and familiarising myself with the corpus and sub-corpora, I began to reflect on its relevance in social terms, offering ways for the researcher to connect with what the data set represents socially and politically. For these reasons, I here explore the idea that a corpus can be seen as an ethnographic site. Specifically, linguistic ethnography is based on ‘reflexivity about the role of the researcher; attention to people’s emic perspectives; sensitivity to in-depth understandings of particular settings; and openness to complexity, contradiction and re-interpretation over time’ (Tusting, Reference Tusting2019: 1). Moreover, one of the directions of linguistic ethnography is digital ethnography, which is ‘interested in ways in which people use language, interact with each other, employ discourses and construct communities, collectives, knowledge and identities through and influenced by digital technologies’ (Varis and Hou, Reference Varis, Hou and Tusting2019: 230). These concepts lie in a solid foundation for discussing the following results, centring the ethnographic aspects around the corpus. Furthermore, the corpus is built from a digital platform, Twitter, a micro-blogging social media platform where public figures, institutions, media, and individuals exchange information. Twitter can be considered as a micro-context, with its own medium-related affordances (mentions, asymmetric following, and hashtags, referred to as social tags by Zappavigna (Reference Zappavigna, Hoffmann and Bublitz2017)), yet also intrinsically linked to and based on what happens, in terms of societal changes, outside the medium – that is, macro-contexts. The corpus compiled for this study is centred around a specific ‘social’ and linguistic event, in this case, the schwa, in what can be defined as an ‘internet event’ (Hine, Reference Hine2000: 50). Similar to an ethnographer, I approached the social media platform Twitter to see what is happening in terms of linguistic behaviour, and especially how. The how is connected to Twitter as a place for the informal exchange of conversations, allowing relations to be forged. This social media site is also described as a semiotic technology (Zappavigna, Reference Zappavigna, Hoffmann and Bublitz2017) in which the high level of possible interactivity is central. In this interactionality, exchange, and sharing, I conceive of my schwa-corpus as a digital ethnographic site where a range of activities, developing both in more traditional and offline traditional ethnographic research, happen. For instance, interactions occur through emojis, mentions, hashtags, and specific forms of community-building, as shown in the analysis. In digital ethnography, time is shifted (Varis and Hou, Reference Varis, Hou and Tusting2019) as the participants and the ethnographer do not have to be there simultaneously. In addition, time can be seen as a fruitful element in the in-depth understanding of the language (and cultural) phenomenon (as described in Section 2.4.1). The corpus thus becomes a specialised archive of how and when the linguistic phenomena occurred. To investigate language, I start from the understanding that ‘social media are part of what can be characterised as the “messy web”, resulting in an equally complex online ethnography process’ (Postill and Pink, Reference Postill and Pink2012: 126). With due differences to other ethnographic research, I am invisible to the speakers who used inclusive language, exploring linguistic actions in digital space and therefore a register (Twitter).

Moreover, I enter a space already inhabited by the language I investigate (see Section 2.1 for information). My aim is to make sense of some patterns having at their core the negotiation of traditional and novel gendered morphemes and, broadly, gender (as per investigation through FCDA). Seeing the corpus as an ethnographic site has been methodologically useful to split the analysis into several parts, some quantitative and based on aspects of frequencies (while also being complemented by qualitative insights), and others purely qualitative on metalanguaging and relevant functions of the schwa-words. Both investigations aim at providing a comprehensive picture of how gender inclusive language is used and the multiple functions it expresses on Twitter.

2.5.1 Corpus-Assisted Research of Gender-Inclusive Language: Quantitative Patterns

With the schwa-corpora at hand, I conducted a collocation analysis of *ə to explore words that statistically co-occur (referred to as collocates in CL) with those ending in the schwa. This allowed me to see whether the schwa was used in isolation or integrated into a more complex syntactical structure consisting of other inclusive terms. There have been criticisms that conceptualise Italian not as a versatile language for the many matching elements (also referred to as gender agreement) that would need changing to be made fully inclusive (see Section 2.3). For this reason, I have collected the collocates in plus two and minus two positions from the node; the rationale for this span is to explore proximate elements such as articles, adjectives, and nouns. Furthermore, tweets are usually brief texts and with the software not recognising sentence boundaries, the risk was that a longer span could capture textual elements from other tweets. The collocates were analysed for the six corpora (the schwa-corpus and each of the five sub-corpora). An MI3 score was used to examine the collocates’ strength based on the observed and expected frequencies. In Table 6, I present the results, divided into the number of collocates with a schwa (for the first thirty words in each corpus), the percentages of the number of collocates that contained schwa-elements, whether noninclusive alternatives were present, and the number of the elements that could have been made inclusive but were not.

From the table, some interesting patterns emerge. Firstly, it shows that having sub-corpora is a good strategy for exploring insights into the distribution and how the schwa has the potential to work differently depending on the linguistic contexts. As can be noticed, the percentage of schwa-collocates ranges from 26.6 per cent for the corpus containing the adjective and pronoun forms to 36.6 per cent for the whole corpus and the Kinship and friendship category. The highest percentage of schwa-collocates can be found in the sub-corpora School (40 per cent), Work (43.3 per cent) and Activism (46.6 per cent). Possibly, the 46.6 per cent in the activism sub-corpus is not surprising because it consists of terms aimed at expressing forms of activism. Interestingly, in the sub-corpus School, there is one collocate which is not used inclusively, yet it is used as a split form (frequency: 7), somewhat indicating the intention of the speakers to avoid a generic term (usually masculine), as can be seen in:

1. (1)

   In bocca al lupo a tuttə.INCL i/le.SPLIT miei/mie.SPLIT compagnə.INCL maturandə.INCL, sopravvivremo anche a questo 🍀 #maturita2021

   Good luck to all my schoolmates graduating, we will survive this too 🍀 hashtag.

This sentence has a mix of inclusive elements – tuttə, compagnə, maturandə – and split forms that are masculine and feminine pairs (see Formato Reference Formato2019) – that is, i.masc/le.fem (the), miei.masc/mie.fem (my).

In exploring the concordance lines of un and del in the Kinship sub-corpus, I found that, on some occasions (and therefore not statistically significant), they were used as (masculine) generics preceding schwa-words, as in (2):

1. (2)

   @mention ti dico la stessa cosa che ho detto a un.MASC altrə.INCL ragazzə.INCL: non so se tu faccia parte della comunità e non mi interessa saperlo, ma se tu non lo trovi irrispettoso non vuol dire che non lo sia dato che altre persone lo hanno trovato offensivo

   I repeat what I said to another young person: not sure if you are part of the community and I have no interest in knowing it, but if you do not find this disrespectful, that does not mean that other people cannot find it offensive.

In this example, the article used is in its masculine form but is followed by a schwa pair, possibly suggesting that some speakers prefer to emphasise some elements rather than others. However, one cannot exclude that speakers found it easier to recognise some elements that can be rendered inclusive and leave some others behind. The topic of articles or preposizioni articolate (i.e., preposition plus articles) is an interesting one, as also dealt with by Slemp et al. (Reference Slemp, Black and Cortiana2020) and Slemp (Reference Slemp2021) when investigating tweets from Spanish. Their research found that speakers use inclusive language, either in the inclusive devices –e (todes) or –x (todxs), preceded by what they call ‘doublets’ (i.e., split forms, e.g., the articles los.masc/las.fem). These initial findings pushed me to investigate the co-occurrence phenomenon more closely. In order to do so, I conducted an n-gram analysis, choosing bigrams (i.e., exploring the frequencies of a sequence of two tokens) of schwa-words belonging to sub-corpora Work, Activism, Kinship and friendship, and School (see Table 4). The analytical framework has been built by manually investigating the bigrams. It explores not only the aforementioned articles and preposizioni articolate but also possessives (e.g., nostr–/our) and determiners (e.g., quest–, this/these), referred to as satellite elements. To summarise, these are the categorisations of the forms:

1. i) Inclusive forms: for example, deə (of the), ə (the), meə (my), questə (this/these)

2. ii) Feminine forms: for example, delle (of the), le (the), mie (my), queste (these)

3. iii) Generic masculines: for example, del (of the), il (the), miei (my), questi (these)

4. iv) Split forms: for example, del/le (of the), il/la (the), miei/mie (my), questi/queste (these), whether they reproduce female firstness (first in the pair is the feminine) or male firstness (first in the pair is masculine)

5. v) Other forms, for example, di (of)

Before explaining the results, I wish to explain some differences between feminine forms and generic masculines. As shown through the examples, it cannot be said that feminine forms are used as generics; they seem to be mismatching, or possibly deliberate use of feminine forms, when compared to the long tradition that has seen masculine forms used regardless of the gender of the referents.

The findings of these forms, divided for each schwa-word in the four sub-corpora, are presented in Table 7. In it, I present the absolute frequencies and percentages representing the relative frequency for each word and total within the sub-corpus and a comprehensive view of the sub-corpora in the final total row.

Starting from the results of each sub-corpus, I calculated the p-value to check whether these results had the probability of occurring randomly; using Pearson’s Chi-squared test (a tool to measure categorical data), the results showed a p-value of 0.0001, demonstrating that these results are statistically significant. Table 7 provides interesting insights into how articles, preposizioni articolate, possessives, and determiners are used with the schwa-words investigated. Firstly, one notices that the total results show that the nouns are preceded by a wide range of inclusive devices (some of which are discussed qualitatively). Specifically, 80.71 per cent of forms are inclusive, showing the versatility and adaptability of forms aimed at achieving gender, or rather, inclusive gender agreement. All terms investigated have inclusive forms in their highest percentage compared to other types (i.e., feminine, masculine, and split forms). Masculine forms are, in the total results, those that follow (15.09 per cent), suggesting that rather than a simply masculine specificity, these still work as generics. This point needs more investigation as (in)consistencies construct interesting scenarios on possible biases about specific terms, as discussed later in this section. Split forms are the third highest used forms, with a small percentage (2.41 per cent). It is perhaps unsurprising that the speakers do not prefer these as they reproduce the binary masculine/feminine (and, with regard to social gender, male and female) and, therefore, they might be seen as hindering the goal of inclusivity. This can be possibly used as a counterargument to Russell’s (2024) scepticism about the possible binarism of gender inclusive forms. Here the speakers seem to understand what is binary and what goes beyond it. There are no cases of split forms that include inclusive forms too – for instance, il/la/lə (or other inclusive forms), and all split forms are gender binary. In addition to this, the split pairs appear in male firstness – that is, as Baker (Reference Baker2013) suggests, when the masculine form precedes the feminine: gli/le or i/le (plural the), dei/lle (of the), il/la (singular the), mio/mia (my), suoi/e (his/her), un/a (a), quei/quelli (those) and sugli/sulle (on the). Female firstness is seen in the pairs alle/agli (at the/to the), la/il (singular the), and le/gli (plural the). In examining the total findings, the lowest percentage is that of feminine forms, with 1.79 per cent. Starting from this, I explain some interesting uses of these forms in the following paragraphs. In relation to inclusive forms, I present Table 8 to show the different forms employed in the tweets.

Previously, I mentioned that inclusive forms are the ones that are mostly used in the corpora. I find this to be an important insight into how speakers find inclusive solutions for gender agreement, as issues around syntactical functioning have always been among the fiercest criticism among linguists (as in Arcangeli’s petition, see Section 2.3) and other speakers. For this reason, I see articles and preposizioni articolate as the speakers’ battle fought on the ground, pushing forward ways in which inclusive language can be fully used. As for articles (the English the), the ə is used in the singular and the plural forms, while other solutions are used together with existing articles – that is: (1) glə replacing masculine plural gli, thus highlighting the efforts made to avoid generic masculines yet drawing from them to create inclusive forms; (2) lə, which resembles the feminine la and the masculine lo, in their singular forms. The indefinite article (the English a/an) is used with three inclusive devices that are the short schwa unə, the 3 possibly replacing the long schwa (з), as per Boschetto’s recommendations (see Section 2.3), un3, and the morpheme x (see Table 1), unx. When we look at the preposizioni articolate (of, at, to, on, in + articles), it can be seen that the schwa, in most cases, replaces the gendered morpheme traditionally used in similar feminine and masculine forms. It might be that the speakers see dell– (of the), all– (from the) or sull– (on the) as general forms, as they are used with singular and plural gendered morphemes. An example of this is della (feminine.sing), dello (masculine.sing), and delle (feminine.plur) replaced by dellə, also encapsulating degli (masculine.plur), as in (3):

1. (3)

   Sono terrorizzata.FEM dalle decisioni che vengono prese per noi e per le future generazioni dietro le porte chiuse dei palazzi istituzionali. Sul futuro nostro e su quello dellə.INCL nostrə.INCL figliə.INCL pende la spada di Damocle della negligenza istituzionale di fronte alla crisi ecologica.

   I am terrified by the decisions made for us and for future generations behind the closed doors of institutional buildings. The sword of Damocles, the institutional negligence in dealing with the ecologic crisis, hangs over our future and that of our children.

In (3), there is agreement among the three elements: the preposizione articolata, the possessive, and the noun, with the solution of dell– plus the schwa. Other prepositions have the same type of inclusive forms – for example, allə, sullə, as in (4) and (5):

1. (4)

   Sostegno allə.INCL attivistə.INCL che si mobilitano oggi, nella giornata mondiale dell’acqua, contro la Gigafactory Tesla a Berlino/Brandenburg!

   (Let’s) Support the activists mobilising today, on World Water Day, against the Tesla Gigafactory in Berlin/Brandenburg!

2. (5)

   Tra i luoghi comuni sullə.INCL attivistə.INCL, uno è particolarmente duro a morire: quello dei.MASC giovani.EPIC. ‘Ci pensano i.MASC giovani.EPIC’, ‘i.MASC giovani.EPIC per il clima’, ‘la battaglia dei.MASC giovani.EPIC’ sono frasi che i media ripetono in continuazione.

   In the common myths about activists, one is hard to die: that of young people. ‘Young people will think about it’, ‘the young people’s support for the climate’, ‘the battle of young people’: media are constantly repeating sentences like these.

However, the investigation of the corpora also shows other types of inclusive forms – for instance, alə (to/at the), delə (of the), and nelə (in the), signalling the efforts of the speakers in avoiding the double ll as that would only cover three out of the five options for the preposition (e.g., dello, della, delle, degli, dei). Example (6) appears as part of a thread (i.e., a chain of tweets) of an association interested in the rights of those who write comics:

1. (6)

   … le prospettive future e le attività di MeFu, ovviamente e necessariamente dal punto di vista delə.INCL autorə.INCL.

   The future work and the activity of MeFu, are seen through the lens of the authors.

Another option employed is the preposition plus the schwa, although it only appears for two prepositions, daə (from the) and aə (at/to the). ‘From the’ is also used in the form də, possibly a more confusing option as it could be both da (from) or di (of), only left to the context to be disambiguated. With reference to the plural form, some other speakers choose to modify the masculine form, using deglə, stemming from degli, as in (7):

1. (7)

   Comincio a renderme conto che i consigli sono il MALE e capisco anche perché non è compito deglə.INCL psicologə.INCL darne.

   I begin to realise that giving advice is BAD and I understand why it’s not the therapist’s task to provide it.

Similar to the variety of options for the prepositions, the possessives and the determiners appear in various forms in the corpus. The determiners seem to follow what happens for the articles, with forms that replace the double ll with one l – for example, quelə – or the use of the double ll – for example, quellə (both translated into English as that/those). Some remove the consonant group and use the root que– plus the schwa – for example, queə – while the English ‘this/that/these/those’ sees the replacement of the gendered morphemes with a schwa in its complete form and shortened versions, questə and stə respectively. Regarding possessives, while speakers do not have problems replacing the gendered binary morphemes with a schwa when there is only one root – for example, nostrə, vostrə – more flexibility is found for other persons. For instance, the first person is used in the forms mieə and mieiə (stemming from miei, the masculine plural), where the latter can be a(n involuntary) typo in writing inclusive from the habitual masculine generic; suoə and tuoə, originating from the masculine plural tuoi and suoi are also used in the same way. In addition to these forms, some speakers used miə/tuə/suə, resembling the feminine plural mie, tue, sue. These results call for a reflection on standardisation as a general term and, more specifically, with regard to the schwa-forms.

One can see that these go beyond the recommendations made by Gheno (Reference Gheno2022) and Boschetto, portraying a more complex scenario in which, perhaps, speakers are experimenting beyond the said and known. For instance, Facchini (Reference Facchini2021) found that using the schwa was not always linear in investigating interpreters. In addition, the multiple forms are not to be seen, as anti-schwa linguists believe, as a weakness. On the contrary, it shows that if these forms are to empower speakers in their relations with themselves and others, then the speakers themselves are to keep control of the linguistic experiment(ation) outside a (recommended) regulatory frame. In other words, speakers engage in (linguistic) identity work and negotiate what they know regarding grammatical gender and how these can be made inclusive. This proves that the idea of languagers and languaging and, in this specific case, genderers and gendering, is central. Genderers, in negotiating inclusive language, are embedding ideas about the standard, which swings between generic masculines and binary forms, and its rejection. As many studies have demonstrated in the past, the standard should not be considered neutral and part of a balanced system, as masculines have for two long considered generics in contrast with the grammatical system containing feminine and masculine morphemes. Therefore, the speakers’ efforts are paramount to understanding how society is changing and how language connects with this. This also raises another important point, as discussed by Banegas and Lopez (Reference Banegas and López2021) regarding gender inclusive forms, that of normalisation within language change. The tweets investigated earlier suggest that schwa-forms are used competently from the speakers’ point of view (pending mistakes or mismatches), and to some extent, one can suggest that the schwa does not create misunderstandings and is used naturally; however, this cannot be seen exclusively through the lens of the speakers, as readers could also be interrogated about naturality in the exchange. I believe the written register of the tweets is an advantage, as those who interact with the schwa-forms might have a chance to reread or become accustomed to the new forms. The reading is not neutral, as some speakers might reject these forms and believe they hinder mutual intelligibility, as Calder (Reference Calder2022) explained through the cislingual perspective adopted by some hearers.

Furthermore, the symbolic political values of grammatical gender and language inclusivity do not exempt from language as a political tool, specifically for far-right parties (see Borba, Reference Borba2019; Reference Borba2022). The tweet that follows does not belong to the corpus; however, it contributes to explaining the scenario in which the schwa appears. Members of far-right parties Fratelli d’Italia and Lega have been observed speaking against inclusive language or mocking it, as in the tweet of former senator Simone Pillon.Footnote ²⁷ In commenting on the new leader of the PD, the first woman in the role, he writes:

I see this in connection with far-right parties defending national borders from foreign threats and, similarly, limiting language. Examples such as these can be traced in the sense of nostalgia for a mythical past, as described in Section 1, triggering what are defined as linguistic guerrilla wars (Cameron, Reference Cameron1995; Borba, Reference Borba2019), as well as constructing moral panic (as explained in Section 1). I now move to discuss feminine and masculine forms as some telling insights show the complexity of thinking inclusive and the various degrees for it. If we explore the sub-corpora, we see that masculine forms are the second preferred form. An exception is the sub-corpus School, where masculine forms ranked third (11.49 per cent), preceded by split forms (13.79 per cent); in the sub-corpus Kinship and friendship, feminine forms, which are usually the least used, appear slightly more used (1.88 per cent) than split forms (1.79 per cent).

The differences are minimal; however, some patterns captured my attention and can be seen as serendipity (see Section 1). For instance, in the sub-corpus Kinship and friendship, the terms ragazzinə and ragazzə, referring to adolescents or young people, behave differently in relation to articles, preposizioni articolate, possessives and determiners. In exploring the concordances of ragazzinə, it can be noted that no feminine forms are employed. At the same time, it is different for ragazzə, possibly hinting at the idea that ragazzə is perceived through a female lens more than ragazzinə, which, in turn, might be seen as a generic. The two terms are mostly used with inclusive forms (70 per cent for ragazzinə and 78.94 per cent for ragazzə) and masculine forms (22.86 per cent and 12.94 per cent, respectively). In the same sub-corpus, two other schwa-words stand out: figliə (daughter/s – son/s – offspring/s) and amicə (friend/s), because they are preceded by masculine forms (11.2 per cent and 23.81 per cent, respectively) as their second preferred choice. Feminine and split forms are numerically distant from the masculine form (feminine: 0.62 per cent for figliə and 0.97 per cent for amicə; split forms: 0.83 per cent and 0.65 per cent, respectively). The plausible explanations concern a cultural bias, perhaps mostly for figliə; Italian is one of those cultures with a legacy of a hierarchy where baby boys used to have a better cultural value than baby girls (see Formato, Reference Formato2019). The masculine forms for amicə could depend on the schwa-word, as it is used with the masculine plural root (amic–, also the root for feminine and masculine singular) rather than the feminine plural root (amich–). In the sub-corpus school, the terms compagnə (schoolmates) and alunnə (students/pupils) also behave differently. Both are preceded mostly by inclusive forms (76.27 per cent and 67.86 per cent); however, compagnə is preceded, in order, by split forms (13.56 per cent), masculine forms (8.47 per cent) and feminine forms (1.69 per cent, only one occurrence). In comparison, alunnə is preceded by masculine forms (17.86 per cent), split forms (14.29 per cent) and no feminine ones. With such small numbers, it is difficult to provide any solid explanation about preferences; however, I believe that some trends can be seen from the concordances shown in Figures 2 and 3.

Figure 2 Concordances of compagnə (sub-corpus school)

Figure 3 Concordances of alunnə (sub-corpus school)

From these concordances, it seems that, unsurprisingly, compagnə is mostly used for high schools/university and from the perspective of the ‘I plus my schoolmates’ (–2, MI3 collocates: mie/my, ex/former, alcune/some, in the pair alcune plus possessive, lə/the, also in the pair lə plus possessive). Differently, alunnə is a more general term for other educational institutions. It is used by people describing pupils/students as a generic category or a specific group (collocating with the definite article lə and the indefinite article unə). There is a suggestion that when inclusive forms are not used or preferred, some generic terms are still seen through the lens of ‘male as norm’ (Formato, Reference Formato2019).

Different trends can be noticed for the term compagnə in a relationship (partner/sub-corpus Kinship and friendships), where the masculine form is second in preference (17.07 per cent) followed by split forms (4.88 per cent) and no feminine forms; Compagnə in the sub-corpus School has as second preferred option split forms (13.56 per cent), followed by masculine forms (8.47 per cent) and feminine ones (1.69 per cent). A different trend can be seen in the sub-corpus Activism when compagnə means comrades, the split forms and masculine forms have the same percentage (5 per cent), and female forms are used too (2.5 per cent). Among these differences, and if we move past the main preference for the inclusive forms, what stands out is the higher use of the masculine form in relationships. Tentatively, one can say that there are still reminiscences of heteronormativity when the word means partner or that, similarly to alunnə (sub-corpus School), this term is considered generic. This shows that there could be different options regarding using masculine forms that can be seen through a social lens.

In addition, it can be noticed that feminine forms are not used for most of the schwa-words, except for colleghə (colleagues). Split forms are not used in relation to libraiə, autorə, disegnatorə, and economistə. The numbers are very small here; however, it seems striking that masculine forms are those employed in the Work sub-corpus (following the preferred inclusive forms). I argue here that a legacy of cultural ideas around gender suitability in the job market could still hover. Some examples are offered here:

1. (8)

   Io ho bisogno di un.MASC psicologə.INCL perché mi rendo conto di non essere in grado di gestire in maniera corretta emozioni e sentimenti

   I need a therapist because I realise that I cannot manage my emotions and feelings.

2. (9)

   @mention Ma poi perché i.MASC scrittorə.INCL etero danno ai personaggi queer sempre finali tragici??

   Also why do heterosexual writers always resort to tragic ends for queer characters??

In (8) and (9), we see masculine forms in both singular and plural forms, showing that, in some cases, efforts to write inclusively are accompanied by old sexist practices. For some words, such as autorə, disegnatorə, psicologə, and dottorə, the inclusive forms originate from masculine roots (autore, disegnatore, psicologo/psicologi, and dottore, feminine forms are autrice, disegnatrice, psicologa/psicologhe, dottoressa, see Section 2.2). Therefore, it is likely that the visible masculine roots for these are still shaping preferences for articles, preposizioni articolate, determiners, and possessives in masculine forms.

To conclude, the analysis of these forms has shown the complexity and flexibility of choices. It has allowed for an in-depth examination of the dynamicity of speakers in building themselves and others through these new forms.

2.5.2 Metalanguaging: Allyship and Rejection

To explore how speakers conceptualise the schwa in sociolinguistic terms, I conducted a corpus-assisted search, searching ‘schwa’ in the schwa-corpus (fifty-three occurrences). I found it interesting how speakers talk about the schwa, which I refer to as metalanguaging, rather than attitudes towards gender inclusive language. The investigation that follows is, precisely, not a systematic attitude study, as the tweets used to provide an overview originate from a corpus-assisted ethnographic work on the corpus. Studies that investigate attitudes are prolific in several languages (e.g., Bonnin and Coronel (Reference Bonnin and Coronel2021) and Slemp et al. (Reference Slemp, Black and Cortiana2020) for Spanish, Nodari (Reference Nodari2022) for Italian, Renström et al. (Reference Renström, Lindqvist and Sendén2022) for Swedish, and Hekanaho (Reference Hekanaho2022) for English) and ad hoc conferences have paid attention to these (e.g., Attitudes towards gender-inclusive language: A multinational perspective, 8–9 September 2022 (an online conference), and Gender-neutral / fair / inclusive / non-binary / non-sexist languages and their dis/contents, 16–17 October 2023, University of Chicago Center in Paris).Footnote ²⁸ Interconnections can be seen between speakers, politics, and gender-inclusive language emphasising the political nature of this linguistic phenomenon. I also covered the debate around novelty and nostalgia, stemming from initial findings from this book, in my talk at Lavender 29 (titled Nostalgia vs novelty: Speakers’ choices resolving criticism towards inclusive language). Moving away from attitudes, Jaworski, Coupland, and Galasinski (Reference Jaworski, Coupland and Galasinski2012) argue that the process of metalanguaging is complex as it touches on several aspects of how communicators refer to, explain, and comment on their talk or that of others. Metalanguage is related to marking ‘personal or group identities’ (Jaworski et al., Reference Jaworski, Coupland and Galasinski2012: 4) and is involved in personal identification work as well as social relationship work. Language is contextualised and explained through representation and evaluation; because of this, metalanguage enters ‘public consciousness’ (Jaworski et al., Reference Jaworski, Coupland and Galasinski2012: 3) and works at ideological levels. In other words, commenting on language is social work. Furthermore, metalanguaging can be associated with the wide debate on speakers (linguists and not) meddling with language through evaluative stances, an aspect that is prominent in the work of Cameron (Reference Cameron1995), the so-called verbal hygiene. In accounting for the space restrictions of this Element, verbal hygiene relates to many aspects of normativity (in relation to grammatical rules), ideology (in relation to linguistic change), social practices (in relationships among speakers), cultural and political significance, and personal linguistic histories (in relation to identity but also agency).

One of the most interesting aspects is allyship and how this is visible through the use of the schwa or talking about the schwa. To investigate allyship, I argue that it appears in indirect and direct forms. The indirect one is described in Sections 2.5.1 and 2.5.2 – that is, using the schwa as a gender inclusive device in nouns and other gendered elements. In my view, direct allyship is when the speaker openly defends and stands by inclusive linguistic devices (the schwa and others). In the following examples, I present some concordances (10) – (14) of direct allyship, each providing nuances in the relationship between the linguistic event and the speaker’s ideas and beliefs around the topic:

1. (10)

   @mention si usa per riferirsi a tutti.MASC. invece di dire appunto ‘tutti’ si dice tuttə.INCL. ad esempio ciao a tuttə.INCL, qualcunə.INCL, ragazzə.INCL ecc

   @mention it is used to refer to everybody, instead of saying, precisely ‘everybody’, one can say everybody, for instance ciao all, someone, youngsters, etc.

2. (11)

   @mention @mention @mention Dottoressa.FEM è il femminile. Dottore.MASC se ti riferisci a un uomo, dottoressa.FEM ad una donna, dottorə.INCL se ti riferisci a un gruppo di uomini e donne o a una persona non binaria. Questa è l’idea dietro allo schwa.

   Dottoressa is feminine. Dottore if you refer to a man, dottoressa to a woman, dottorə.INCL if you refer to a group of men and women or to a non-binary person. This is the idea behind the schwa.

3. (12)

   @mention Allora io lotto ogni cazzo di giorno così come lə.INCL mie.FEM compagnə.INCL. Da anni. A te non serve la schwa perché non sei una soggettività esclusa. Serve TUTTO. Una cosa non esclude l’altra. Se non la vuoi usare non la usi, ma per moltə.INCL è un passo importantissimo.

   So, I have been fucking fighting every single day as much as my comrades. For a long time. You do not need (to use) the schwa as you are not excluded as an individual. Everything is needed. One aspect does not exclude the other. If you do not want to use it, don’t but for many, this is a very important step.

4. (13)

   avevo già iniziato ad usare lo schwa quando ci scrivevamo, e nei vocali evitavo di usare aggettivi riferiti alla sua persona per non declinarli al maschile o al femminile.

   I had already started to use the schwa when we were writing to each other and in audio messages to avoid using feminine or masculine adjectives.

5. (14)

   nb: nel thread ho usato la schwa per parlare di lesbiche perché he/him e they/them lesbians esistono e in particolare alcunə.INCL miə.INCL mutuals lesbichə.INCL usano i pronomi they/them e non voglio rischiare di dare fastidio a qualcunə.INCL, vi prego di fare lo stesso.

   Please note: in the thread I used the schwa to talk about lesbians because he/him and they/them lesbians exist and specifically some of my mutual lesbians use the they/them pronoun and I do not want to upset anyone, so I ask you to do the same.

In (10) and (11), the speakers suggest how the schwa should be used and who it refers to, offering some examples. In (12), the tweet is an answer to someone who is more likely opposing the use of the schwa (the original tweet was deleted); the speaker fiercely explains how the schwa resolves the feeling of exclusion in society and language, reproducing empathy (Kolek, Reference Kolek2022). Whether or not the speaker forms part of this group, as it remains unclear from the tweet, the allyship is seen in the alignment to the schwa as a symbolic resource for many (per moltə). In connection with the public debate, this is an interesting and telling point as it explains the political force embedded in the morpheme, contrasting those views that see the schwa as an imposition. Rather, we see in these examples and in others in this section that gender inclusive language can be seen as a site of advocacy (Konnelly et al., Reference Konnelly, Bjorkman and Airton2022). In my view, the discourse of feeling compelled to use this device is used purposefully and politically to reject ideas around inclusive language and gender, fuelling moral panic. In (13), we see a speaker explaining their choices with respect to a specific individual. This tweet can be seen through the notion of outgroup-focused motivation, the desire to improve the condition of a discriminated and disadvantaged group (Radke et al., Reference Radke, Kutlaca, Siem, Wright and Becker2020), in which one’s own privilege is recognised. Similarly, in (14), an elaboration of how the schwa is doing allyship work for a specific group of people, some of whom are in the speaker’s social circle. Differently from (12), example (14) shows the speaker asking for collaboration on how to represent other people linguistically, inviting a reflection on respect (vi prego di fare lo stesso/I ask you to do the same), which can be seen as a moral motivation (Radke et al. Reference Radke, Kutlaca, Siem, Wright and Becker2020). For them, morality is based on an identification with a superordinate group and action is prompted by the recognition of the disadvantages suffered by the group. From the same perspective, in (15), there is a positive message in relation to the schwa:

1. (15)

   Comunque ho letto più post con la schwa in queste 24 ore che nei 24 mesi precedenti. E appena sarà presente come carattere in tutte le tastiere la vedremo crescere esponenzialmente. Perché le idee non si fermano, ragazzə.INCL

   I read more posts with the schwa in the last 24 hours than in the previous 24 months. And as soon as it will be available on keyboards, we will see it exponentially more. Because we cannot stop ideas (from blossoming), folks.

No clues can help categorise allyship work, as this seems a more general statement about the observed use of the schwa. However, reading the speaker’s alignment with the cause is possible. Similarly, speakers are also willing to engage in sociolinguistic discussion of the phenomenon through a positive lens as, for instance, in (16):

1. (16)

   @mention esiste anche la schwa [ə] (al plurale [ɜ]) che si pronuncia come la fine delle parole in napoletano! è neutro e perfettamente pronunciabile, e si usa così: oggi sono andatə.INCL a fare la spesa con lə.INCL miə.INCL migliore amicə.INCL. poi siamo andatɜ.INCL al parco. ci siamo divertitɜ.INCL tantissimo!

   @mention there is also the schwa [ə] (plural [ɜ]) that is pronounced like the ending in some Neapolitan words! It is neuter, perfectly pronounceable, and used as follows: Today I went shopping with my best friends. Then we went to the park. We enjoyed it a lot!

This tweet is sent in response to a thread that deals with non-binary linguistic preferences, where information about personal subject pronouns is also shared. Similar to the other cases discussed in Section 2.5.1, the speaker here explains how the schwa can be flexible when matching all elements in the sentence, following Boschetto’s recommendations on the use of the long schwa. The reference to the Neapolitan dialect is useful for understanding the Italian context and is employed to explain how this sound is not alien or foreign. In Section 2.3, I suggested that, conversely, linguists and the general audience use this geographical-oriented argument to demonise both the Neapolitan dialect and the gender inclusive schwa. More specifically, in the change.org petition, the linguist Arcangeli writes that the schwa is peculiare di diversi dialetti italiani, …, stante la limitazione posta al suo utilizzo (la posizione finale), trasformerebbe l’intera penisola, se lo adottassimo, in una terra di mezzo compresa pressappoco fra l’Abruzzo, il Lazio meridionale e il calabrese dell’area di Cosenza (is specific to several Italian dialects, …, in its limited use as the ending of a word, if we had to adopt it would turn the whole peninsula in a middle earth in between Abruzzo, the low Lazio and the Calabria area of Cosenza).Footnote ²⁹ This linguist describes an area that is bigger than where Neapolitan is thought to be spoken, including the low Lazio, the region bordering Campania in the north, Abruzzo (bordering Lazio from the west) and Calabria, which is South of Campania. In his argument, there seems to be more than a neutral geographical description of the sound schwa. I believe that a discriminatory attitude towards these southern areas can be read through the lines. It is not novel to find discrimination with reference to southern dialects, stemming from a broader and long-term social discrimination based on stereotypes about education, engagement with the state, and criminality (mafia) (Mioni and Arnuzzo-Lanszweert, Reference Mioni and Arnuzzo-Lanszweert1979).Footnote ³⁰ The north–south divide is historical and goes back to the unification of Italy in 1861. Furthermore, in (16), there is also what I believe is an interesting argument: the idea of the schwa being neuter. In this Element, I have widely discussed that terms such as ‘neutralisation’, ‘neuter’, and ‘neutral’ are possibly a disservice to what gender-inclusive language does, which is not to cancel, erase, and neutralise gender. It does the opposite, in my view, as it makes gender visible and political. It, specifically, tells hearers about discrimination, injustice, and the different future one wishes or expects; it is part, in other words, of the construction of a more welcoming society. In terms of linguistic realisations, speakers who are allies seem to have clear what is still central to the debate of both gendered and gender inclusive language, as shown in (17):

1. (17)

   Il problema non è lo schwa, la difficoltà nel pronunciarlo, ‘l’inesistenza in italiano’(?) o l’Accademia della Crusca che non vuole. Il punto è che il maschile sovraesteso piace perché – secondo alcunə.INCL – i ruoli di decisione devono essere ricoperti da uomini.

   The schwa is not the problem, the trouble in pronouncing it, ‘the nonexistence in Italian’(?) or the academy of La Crusca that does not want it. The point is that the generic masculine is liked because – according to some – men must hold decision-making roles.

Here, we see how speakers conceptualise the schwa regarding broad ideas about language and society, foregrounding a mystification put forward by those who reject this inclusive device. Genderers are, therefore, moving within a space where there is a constant renegotiation of linguistic forms, where they acknowledge that sexist generic masculines continue to exist and be meaningful in a male-oriented society (Formato, Reference Formato, Venuti, Vigo and Campisiforthcoming).

Through these examples, we have seen how direct allyship can raise awareness of language functioning, highlighting activisms by individuals helping shape the debate in several ways. Allyship also needs to be seen in relation to its counterargument – that is, the firm rejection of these forms. For instance, some resort to the known explanation that there are more important things than language (see Formato, Reference Formato2019; Vergoossen, Pärnamets, Renström, and Gustafsson Sendén, Reference Vergoossen, Pärnamets, Renström and Gustafsson Sendén2020), as shown in (18):

1. (18)

   @mention @mention Ma secondo voi davvero studentesse.FEM e operaiə.INCL, docenti.EPIC e attivistə.INCL discutono insieme da anni nelle assemblee dello schwa? Ma voi davvero fate?

   Do you really think that students, factory workers, teachers, and activists debate the schwa in their meetings? Are you for real?

While not directly showing why the schwa is an unsuitable solution, this tweet narrates a specific story concerning job roles or activism. More specifically, it can be seen through the lens of social class – for instance, the mention of operaiə (factory workers), but also arguably low-middle class professions such as docenti (teachers), as well as female students (studentesse, seen as a homogeneous gendered group) – points at the schwa as an empty effort emerging from so-called salotti femministi. This expression frequently suggests that useless fights happen in higher social class circles, created by and for those who do not have to face class discrimination. I see this as one of the serendipity moments (as described by Partington et al., see Section 1.3.1), as it provides an opportunity to reflect on intersections and intersectionality. The tweet is written by a man whose bio suggests he is a professor at an Italian university, casting doubts that this is a personal experience (for gender and, arguably, class). It is more likely an attempt to dismiss the relevance of gender inclusive language, emphasising a class struggle to undermine feminist efforts and its relation to the working-class. While more research is needed on how allyship works, possibly rethinking the categories, I argue that this is an important lens through which speakers participate in the exchange of ideas and values on Twitter.

2.5.3 Exploring the Corpus Ethnographically: Functions of Gender Inclusive Language

In this subsection of the analysis, I report on uses of the schwa that refer to social functions, where social refers to both how society is perceived and valued in linguistic choices as well as the medium in which these functions occur (Twitter). I unravel how the schwa is deliberately employed to convey several aspects of interactional work on Twitter through a corpus-assisted exploration of concordance lines that appeared while conducting the aforementioned investigations. In other words, these are qualitative notes that I made along the way, similar to those of an ethnographer that explores what happens in a site. While cherry-picking elements cannot be excluded, as for the discussion in the previous analysis, these are to be considered as an illustration of plausible functions in this corpus. They must not be considered the only functions, as others might emerge in different corpora. The more obvious function that is found in the corpus is that of self-representation. That can be seen as a way for people to, perhaps indirectly, tell their gender story, as in ((19)–(21)):

1. (19)

   In questo scenario io sono l’amicə.INCL che parla della serie. GIF

   In this case, I am the friend who talks about the TV series. GIF

2. (20)

   Ti sbagli! Mi sono trasferitə.INCL in Lussemburgo dopo il master!

   You’re mistaken. I moved to Luxembourg after the MA programme!

3. (21)

   E pensare che dopo questo mi sono fidanzatə.INCL e ho trovato lavoro senza volerlo GIF

   And right after this I found a partner and a job without looking for them. GIF

In (19) – (21), the individuals posting on Twitter use the schwa to talk about themselves, suggesting they do not wish to be identified with feminine or masculine morphemes. Using the schwa can be seen as different while sharing some elements of more traditional coming out stories. In (19), there are mismatches – for example, the article preceding amicə, which can be a reference to the binary, whether feminine (stemming from l’amica) or masculine (l’amico). Perhaps because of this versatility, it is seen as inclusive, not having a gendered morpheme visible. On other occasions, the schwa is used to construct specific communities to which individuals participate or belong, as in the case of LGBTQIA+ ones, of which (22) and (23) are examples:

1. (22)

   BUON PRIDE MONTH A TUTTƏ.INCL!

   Happy Pride Month to all!

2. (23)

   Siamo amicə.INCL, colleghə.INCL, vicinə.INCL di casa, genitori.EPIC, figliə.INCL che chiedono a gran voce di essere riconosciutə.INCL e tutelatə.INCL 👉 milanopride.it

   We are friends, colleagues, neighbours, parents, children who are asking, in a loud voice, to be recognised and legally protected 👉 milanopride.it

In these examples, the schwa takes an added meaning as it constructs a specific group of people. In (22) and (23), the groups mentioned through the schwa seem to represent various identities in the gender and sexuality spectrum, as seen with reference to the pride event in Milan. It is possibly unsurprising that the identification with these groups is made visible through the schwa.

On the subject of political allyship, the mainstream left-wing PD has never openly spoken about gender inclusive language, and its website carries traces of generic masculines – for example, i deputati, i senatori, gli eurodeputati (the MPs and MEPs). Conversely, a smaller political party, arguably not mainstream, Possibile, uses the schwa to signal its stance and as a direct allyship tool (see Section 2.5.2). In (24), I present an example:

1. (24)

   Bologna Possibile partecipa alla piazza #MOLTOPIUDIZAN prevista per questo sabato alle ore 12:00 insieme a tuttə.INCL lə.INCL compagnə.INCL del #RivoltaPride contro l’affossamento del #DDLZan al Senato.

   Bologna Possibile takes part in the #moltopiudizan mobilisation, on Saturday at 12 together with all comrades of #RivoltaPride against the collapse of the #DDLZan in the Upper Chamber.

In this example, the main inclusive term is compagnə (comrades), reproducing a linguistic item associated with left-wing parties in the past. The novelty of the inclusive device further suggests where the party stands in relation to the LGBTQIA+ community. The choice of the schwa is here related to how the political party aims to connect with the Twitter audience. Similarly, Burnett and Pozniak (Reference Burnett and Pozniak2021) found that the écriture inclusif is used as a political tool, where French universities with a left-wing activism seemed more inclined to use the point médian (while those aligning with the right used binary forms, e.g., parentheses). In Formato and Somma (Reference Formato and Somma2023), we also explore how inclusive language can be used in or be meaningful in sociolinguistic communities, specifically Communities of Practice, Speech Communities, and Imagined Communities. The renegotiation of gendered terms touches on several aspects of traditional uses of masculine generics. For instance, in Formato (Reference Formato2019: 69), I noticed that some terms used in their masculine forms indicated a role rather than a person. While still anchored to a history that saw men suitable for specific professions, personalised masculines would go further than addressing or describing a woman in a traditionally male job. In other words, these terms would become a representation of the job or the professional category (e.g., per essere il sindaco.masc/to be a mayor). Speakers, through the schwa, are also able to renegotiate this view, as in (25):

1. (25)

   Nel futuro non ci servono così tanti.MASC ingegnerə.INCL. Educatorə.INCL, medicə.INCL, infermierə.INCL, psicologə.INCL, caretakers in generale, artistə.INCL ecc.. sono queste le figure di cui avremo più bisogno

   In the future, we won’t need so many engineers. Educators, doctors, nurses, therapists, caretakers, broadly speaking, artists, etc … are the professions we would need the most.

Replacing these masculine forms with inclusive ones shows how speakers envision a society that is more inclusive and different from a known gendered order regarding the suitability of gendered workforces. However, tanti (many) is still used in its masculine form, demonstrating that mismatches still occur (see Section 2.5.1).

In terms of how the schwa collaborates with other linguistic devices, the corpus offers some examples, such as the use of mixed inclusive forms, as in (26):

1. (26)

   Ah quindi è stato assunto per fare il.MASC baby sitter a*.INCL figliə.INCL di qualcun*.INCL.

   Ok, this means he has been hired to babysit someone’s child.

In (26), there are three inclusive elements: the preposizione articolata (a*), the noun (figliə), and the indefinite pronoun (qualcun*). It is difficult to interpret why the speaker has decided to alternate the inclusive forms; perhaps it was easier for the preposizione articolata, while qualcun– would have worked similarly with both the schwa and the asterisk. I perceive this as a sign of flexibility that could be useful for the speakers to have their message on where they stand to come across (as also seen in Slemp, 2020 and Burnett and Pozniak, Reference Burnett and Pozniak2021).

Overcoming the personalised masculine (as in (25)) and the mixed form (as in (26)) is part, in my view, of how the schwa could become, for some speakers, a new generic, de facto substituting the sexist masculine one, which is fairly seen as a loaded legacy of the past. This is not to say that other strategies do not have the same potential on their own or in coexistence with others, and other strategies might as well emerge in the future. Burnett and Pozniak (Reference Burnett and Pozniak2021: 810), starting from their investigation of the point médian in French, similarly believe that gender inclusive language has the potential of ‘eliminating, or at least reducing, linguistic androcentrism’.

A confident competence is also found when speakers are called to reproduce gender and specific referents, as in (27):

1. (27)

   Io, mio.MASC padre.MASC, i.MASC miei.MASC fratelli.MASC, le.FEM mie.FEM nonne.FEM, la.FEM mia.FEM amica.FEM, lə.INCL miə.INCL amicə.INCL, gente su Facebook a cui ho chiesto in un gruppo. GIF

   Myself, my father, my brothers, my grandmothers, my friend, my friends, people on Facebook whom I contacted through a group. GIF

In this example, it is clear that the speaker is in control of the representation of gender in linguistic terms, showing that gender inclusive language does not erase women. The range of participants referred to in this tweet shows a certain flexibility in using gender inclusive language and in signalling the gender of the specific people (mio padre, i miei fratelli, le mie nonne) and groups (lə miə amicə). Here, gente (people) follows the idea of language neutralisation in constructing groups where gender is not made visible (cfr, the debate on terminology in Sections 1 and 2.3).

Another aspect found while exploring the corpus is the replacement of idiomatic forms through the inclusive schwa. With idiomatic expressions, I intend those usually understood as conventional; the speakers recognise the expression as familiar and having a specific meaning. In this case, the idiom I found is chiedo per unə amicə (asking for a friend); this is a very common expression used on social media to pretend that the question asked does not stem from or concern the person asking, who instead becomes an intermediate for someone else – that is, an unnamed friend. Having checked the corpus, the grammatical gender in the idiomatic expression sometimes matches that of the speaker – for example, chiedo per un’amica for a female speaker. At the same time, in other cases, the masculine or masculine generic form, chiedo per un amico, is used. The willingness to (re)create groupness and connectedness through a request for interaction is core in social media. From this perspective, Leppänen et al. (Reference Leppänen, Kytölä, Jousmäki, Peuronen, Westinen, Seargeant and Tagg2014: 113) describe social media as an ‘affinity space’ where what occurs originates from ‘shared interests, causes, lifestyles, activities and cultural products with short life spans or passing popularity’. The timeliness is central to the aforementioned expression (chiedo per unə amicə), which can be seen as a ‘cultural product’ that could work in face-to-face interactions too. The expression is, therefore, a search for alignment in the togetherness of communities on Twitter. Regarding alignment, Zappavigna (Reference Zappavigna, Seargeant and Tagg2014: 139) explains that microblogging (the practice of posting on social media) revolves around ‘share quotidian experience by conferring upon the private realm of daily experience a public audience’. This fits with what chiedo per unə amicə does, with the quest to start a conversation. Similarly, I find the idea of phatic communion, reframed by Zappavigna (Reference Zappavigna, Seargeant and Tagg2014) in the social media context, interesting: communication is centred around establishing bonds rather than merely or exclusively communicating ideas. Furthermore, the notion of a phatic media culture is telling, where bonds and sociability are the main objectives in a wide network (worldwide) (Zappavigna, Reference Zappavigna, Seargeant and Tagg2014). More significant, in the context of social media, posts by individuals are an ‘I am here’ declaration, what Makice (Reference Makice2009) defines as ‘linguistic ping’. For this reason, I argue that using the inclusive form, chiedo per unə amicə, is an explicit way to frame the individual’s choices behind the ‘I am here’, one that supports the vision behind the inclusive device.

In exploring the term carə through concordance lines and manual analysis, I noticed that it was employed to refer to different groups involved in less or more proximity to the speaker. This is not to say that the forms cara.fem.sing, care.fem.plur, caro.masc.sing, cari.masc.plur (dear) are not used symmetrically to carə. Carə, in this corpus, was found to address the internet community and specific people with whom the speakers wish to interact through the mention function (@userhandle). On some occasions, carə was used to mean someone’s loved ones or in a personal form (carə me/dear me). However, in addressing the internet community, we have two additional functions, which are (a) addressing the whole community to build a similar bond to that in (27), shown in example (28), or (b) addressing the whole community to advertise a product or an event, as in (29), sharing information on an event on data mining.

1. (28)

   carə.INCL, vorrei andare anche io al mare …

   dear, I would like to go to the seaside too …

2. (29)

   Amicə.INCL carə.INCL, a giorni torniamo da voi con l’indirizzo esatto, e anche qualche suggerimento sulle zone di Catania.

   Dear friends, we will soon inform you of the right address and some recommendations about Catania.

What emerges from these tweets is the interactional nature of the discussion on Twitter, the search to build communities with strangers and known people. This is a specific feature of the schwa on Twitter: the written register and the multiple options for the speakers to recreate the welcoming space embedded in the use of the schwa. Similarly to previous functions, I also captured the use of the schwa for humour:

1. (30)

   Cosa dice unə.INCL contadinə.INCL a suə.INCL figliə.INCL quando fa battute idiote sul campo? Piantala

   What does a farmer say to their children when they make stupid jokes in the fields? Plant it!

This joke does not work in English, as piantala in Italian means both ‘plant it’ and ‘stop it’. I did not recognise this joke as conventional, nor did it appear on Google, which I consulted to check whether its standard version was in the masculine form, as other conventional jokes. This means that the power of the schwa is pervasive; it enters different domains of communicative events and functions.

To sum up, the functions of the schwa observed in the corpus are as follows: self-representation, representation of the self in a wider group, construction of specific and internet communities, flexibility in alternating gender inclusive devices as well as in making gender visible in relation to different referents, social media idioms and jokes. This proves that creativity is at the core of this linguistic experiment, despite the criticisms of the schwa being limited and limiting its potential in substituting traditionally gendered morphemes.