The use of gender-neuter marking in Brazilian Portuguese tweets: a sociolinguist approach 1

: This research investigates gender-neuter marking in Brazilian Portuguese, a recent linguistic variation that has divided many opinions in Brazilian social and political circles until now. The Theory of Variation (LABOV, 2008 [1972]; WEINREICH; LABOV; HERZOG, 2006 [1968]), with a focus on Communities of Practice (ECKERT; MCCONNELL-GINET, 1992a, 1992b), supports the description and analysis conducted. To observe the variant, we opted for the social network Twitter, from which we collected 3,446 tweets produced by three subjects during the second half of 2020 who explicitly expressed a preference for the use of the neuter pronouns “Elu,” “Ilu,” “El” or “Ile.” As a selection criterion, we considered tweets containing adjectives, nouns, and pronouns whose referent was a human, and morphological markers carrying gender distinction. The predictor variables stipulated were morphosyntactic class, lexical item, the tweet topic, participant, and the reference performed by gender marks (generic or speci ﬁ c). The sample was submitted through the RStudio program, version 4.1.2, through which the proposed statistical treatment was carried out. The results reveal a disadvantage regarding gender-neuter marking for all participants, not showing optimistic estimates about any speci ﬁ c word. They also pointed out the adjective class, the a ﬀ ective relationship and identity topics, and the generic reference performed by gender marking as conditioners of gender-neuter employment.


Introduction
The normalized conception that gender, biological sex, and sexual orientation are restricted to the binary and heteronormative field, what Bernini (2011) calls the sexual binary system, is a kind of logical operator that constitutes the notion of sexual identity in Western societies.This notion imposes two poles for sex (male or female from the biological point of view), for gender (male or female from the social point of view), and for sexual orientation (heteronormativity from the sexual point of view).Consequently, gender identities not limited to a binary notion are situated in a marginalized position, liable to repression and social estrangement for differing from the standard established by the sexual binary system.This binary perspective is intrinsic to the modern worldview and way of thinking about issues related to sexuality, which has, as a consequence, as Bernini (2011, p. 20) points out, "a macho and heterosexist hierarchy that assigns heterosexual men the status of majority identity, and the other identities resulting from the composition of these terms [...] the status of moral minorities" 2 .
In recent decades, linguistic movements driven by speakers of various languages such as English (XIE, 2015;GALUPO et al., 2017;ELLIOTT, 2017), Swedish (TAVITS; PEREZ, 2019), and Spanish (UBA Sociales, 2019) emerged to promote the inclusion of people who do not fit into the categories of female or male through the establishment of a thirdgender category, the non-binary gender, which came into use as an umbrella term to contemplate individuals whose identities are not fully defined as feminine or masculine.As Reis and Pinho (2016, p. 14) explain, individuals contemplated by non-binary gender identity "will not be exclusively and totally woman or exclusively and totally man, but [...] will permeate in different forms of neuterity, ambiguity, multiplicity, partiality, genderless, otherness, fluidity in their identifications." As a moral minority, individuals of non-binary gender identity create, through verbal language, symbolic resources that determine opposition to the sexual binary system values.In Brazil, the movement for a more inclusive language follows the advance of the use of social networks, occurring, at first, from the registration of the characters "@" and "x" in place of the gender suffix and, more recently, from the use of the "-e" mark: an alternative of pronounceable neuter marking more adopted in social networks today.In this sense, we can observe an effort of the non-binary and LGBTQIA+ community, in general, to inform and explain the functioning of 2 Free English to Brazilian Portuguese translation to all direct quotations.gender-neuter marking through posts, mainly on social networks, which characterizes a movement of appropriation of the rules of the language that goes against the grammatical norm of Portuguese.
Thus, this research, which emerges from the social need to discuss linguistic variation and gender-neuter markers from a scientific perspective, has as its primary objective to describe the use of these markers in the social network Twitter, which has proven to be a space for discussion and use regarding the neuter language.From the theoretical and methodological assumptions of Variation Theory (LABOV, 2008(LABOV, [1972]]; WEINREICH; LABOV; HERZOG, 2006HERZOG, [1968]]) and Communities of Practice (ECKERT; MCCONNELL- GINET, 1992aGINET, , 1992b) -sociolinguistic approaches that observe language in use -this study seeks to identify the conditioning role of linguistic variables of gender-neuter use on Twitter by observing them as resources for the construction of individual and group identities.
Regarding the next sections, we present the functioning of the gender suffix in Portuguese in section 2, and the concept of language that underlies this study, precisely the one adopted by the Theory of Linguistic Variation and Change (LABOV, 1972) in section 3. Section 4 deals with the methodological procedures adopted, with details on the delimitation of the sample and the predictor variables considered, and, in Section 5, we present and analyze the results obtained.Finally, a conclusion section and references follow this section.

Gender Suffix in the Portuguese Language
According to Camara Jr (1970), there is only one gender inflection in the Portuguese language, which is the addition of the inflectional suffix "a" to mark feminine forms, such as "lob(o)" + a, which results in "loba", or "autor" + a, which results in "autora." 3he masculine form is taken as the unmarked gender, used in generic reference, and thus employed due to the lack of a grammatical resource in the Portuguese language capable of contemplating the human gender.According to Endruschat (2015, p. 312), this means that, in Portuguese, "masculine nouns have a wider lexical and referential potential and may be used to refer to males, to groups of people whose gender is unknown or unimportant, or even to female referents."Thus, "generic masculines may evoke male images and can, therefore, be interpreted as androcentric," (ENDRUSCHAT, 2015, p. 312) with its use associated with male dominance in the real world.The impression that masculine gender is dominant in Portuguese is further reinforced by the small group of Portuguese words of trait [+sexed] (in which gender marking tends to be semantic), formed predominantly by masculine forms, given the cognitive salience related to gendered words (SCHWINDT, 2020a).
The two aspects pointed out above, (i) the notion of masculine as an unmarked gender and (ii) the cognitive salience related to gendered words, can be considered elementary to understand the factors that motivate the gender-neuter use since they contribute to invisibilize other genders.As a resource of ideological expression, gender-neuter represents a position contrary to the values implicit in using the masculine gender, which explains why individuals of non-binary gender identity generally avoid the masculine gender, despite it being morphologically neuter.Schwindt (2020b) distinguishes proposals for inclusive language use in Brazilian Portuguese as (1) the use of the feminine gender in two-gender common nouns (e.g., "presidenta"); (2) the use of both feminine and masculine forms instead of generic masculine use (e.g., "alunas e alunos," "todas e todos"); (3) the inclusion of the characters "x" and "@" at the end of nouns and adjectives (e.g., "amigx," "amig@"); (4) the extension of the function of already existing marks (such as "-e" in "amigue") and ( 5) changes in the base or root of pronouns and articles (e.g., "ile," "nile," "dile," "aquile," "le").
Cases (1) and ( 2) are identified as a proposal to highlight the feminine gender through the contrast between feminine and masculine to serve as the binary-gender affirmation in the gender system (SCHWINDT, 2020b).In contrast, (3) and ( 4) are considered strategies of gender neutralization4 (Trubetzkoy, 1939), through a third mark beside the masculine and the feminine.However, in the case of "@" and "x," they have no correspondence in the Portuguese phonological system because they are unpronounceable.The vowel "-e" as a neuter gender mark appears precisely as a possible solution to this problem, being a pronounceable alternative that can be used in oral language and already exists in Portuguese, despite implying "a thorny process" (FREITAS, 2015, p. 167) since the adherence to the neuter gender causes the alteration of several grammatical items to achieve the proper agreement.Schwindt (2020b) emphasizes that gender, more than a piece of lexical information, is a grammatical mechanism, which, in the case of neutralization by "@," "x," or "-e," can cause complications related to the determining articles and pronouns, as well as the pronominal retaking, as exemplified by "Meus(?) dois(?) amigues mais próximes, Vini e Léo, chegaram.Preciso dar atenção a eles(?)"5 (SCHWINDT, 2020b, p. 17).In this example, the neutralization of the determiners and the pronoun in the anaphoric function proves to be a complex process, which is not solved just by adding the vowel "-e" but depends on formal learning to some extent.Additionally, Schwindt (2020b) points out that the employment of the characters "@," "x," or "-e" as neuterizers in the definition of a semantic reference can be an issue when considering which referents are contemplated by these uses.The representative focus of these neuterizers is not only on individuals whose gender identity lies on the non-binary spectrum but can also contemplate groups formed by people of mixed and indeterminate gender.According to Schwindt (2020b), at least two uses are observed for these markings: a generic one for male and female reference and one designating a third gender, as the author exemplifies (SCHWINDT, 2020b, p. 16) in "a) Amigues querides6 , I use "-e" because I do not want to exclude anyone; b) Bom dia amigas, amigos e amigues!7 ".In these cases, a distinction is made between "-e" as a generalization strategy for a group of people of mixed/indeterminate gender and "-e" as marking a third gender, highlighting a group of people of non-binary gender identity, similar to the strategies mentioned earlier for promoting female visibility in discourse.An overview of the employment of gender neutrality in Portuguese is offered by Mokwa (2019) and Almeida (2020), according to Table 1 below.According to the descriptions presented by the authors, based on online guides 8 organized by the non-binary community, it can be observed that there are currently four neuter gender systems in Portuguese, and adherence to one or the other is still a matter of debate in the non-binary community.As compiled by Mokwa (2019) and Almeida (2020), the four most recurrent possibilities of neuter marking in LP (Table 1) are a kind of adaptation of the neuter demonstrative pronoun in the nominative form "illud" 10 of Latin origin.The "Elu" system, closest to the binary pronominal form in Portuguese, is characterized by the use of "-u" as a gender-neuter mark in place of "-a" or "-o" and is used in nominative, third-person possessive and demonstrative pronouns.In the "Ilu" and "Ile" systems, part of the original Latin word ("illud") is retained, with "-u" being the gender-marking morpheme in the former.From this perspective, the "Ilu" system resembles the "Elu" system in using the gender morpheme, di ering in the word's root.In the "Ile" and "El" systems, on the other hand, there is no speci c gender morpheme, so a distinction is made between feminine and masculine in the former because of the Latin root, which preserves the "i" instead of the "e" of "he," and in the latter by excluding the gender morpheme so that in nominative, third-person possessive and demonstrative pronouns, the word ends only with "l."For rst and second-person possessive pronouns, as well as for articles, the forms "mi/minhe," "su/sue," and "le" are adopted by all systems.However, there is no consensus in the rst two cases regarding which 9 Table subtitles: Ele/ela -he/she; dele/dela -his/her; meu/minha -my; seu/sua -your; aquele/aquela -that; artigo o/a -article the. 10 The feminine and masculine forms of this same case in Latin are "illa" and "ille," which refer respectively to the pronouns "aquela" and "aquele" in Portuguese.
competing forms would be more appropriate or which linguistic context favors one usage.

Theory of Variation and Linguistic Change
According to Weinreich, Labov, and Herzog (1968), a linguistic change begins when one of the features of variation spreads through a specific subgroup of the speech community, takes on the character of ordered differentiation, and may symbolize social values attached to that group.In this sense, the adoption of gender-neuter marking becomes a means to, as Santana (2021, p. 708) states, "stimulate debates with the power to destabilize already established meanings and point to the diversity of references with which new meanings [...] can be constructed by subjects".
From this perspective, it is understood that the neuter gender marking acts, within the referred community, as a phenomenon of linguistic variation, considering that the process of variation involves the transfer of innovative linguistic traits from one speaker to another in a community, characterized by the existence of two or more simultaneous variants until a possible change takes place and makes one of the forms obsolete (WEINREICH;LABOV;HERZOG, 1968).This notion of community can be associated with communities of practice, through which social meanings, symbolic values, and identities are constantly and mutually constructed (ECKERT; MCCONNELL-GINET, 1992b), and the use of gender-neuter marking may be seen as an act that represents the values shared by this community.From this perspective, it is possible to understand linguistic variation as a resource for expressing identities, through which information, values, and common learning are shared.Social networks like Twitter, in this sense, can play a facilitating role for the organization of communities of practice, as a space where people belonging to the LGBTQIA+ community (mainly those who identify as non-binary) can interact, joining in for common interests, as seeking for more visibility and respect.In this context, the use of the gender-neuter marking can be seen as a fight strategy (SANTANA, 2021): a linguistic use related to the community through which new meanings are established and non-binary identities are expressed.
Gender-neuter marking occurs, in this sense, in alternation with other grammatical genders in two contexts: (1) in opposition to feminine and masculine markings when referring to subjects of non-binary gender identity, and (2) in opposition to masculine marking when referring to groups composed of people of unknown or varied gender.It is observed, therefore, the coexistence of variants that can be used in the same contexts alternately but whose choice represents di erent social values, associated or not with the community of practice that includes people of non-binary gender identity.
Therefore, the Theory of Linguistic Variation (LABOV, 1972) is used as a basis for describing and explaining gender-neuter marking in tweets produced in Portuguese.This theory represents a theoretical-methodological model developed in the 1960s based on the notion of ordered heterogeneity (WEINREICH; LABOV; HERZOG, 1968), according to which the variation found in languages is conditioned by both linguistic and social elements, subject to identi cation.

Sample
The sample consists of tweets -texts of up to 280 characters from the social network Twitter -known as a microblog.The platform has been characterized by providing dialogue in various elds of human activity since its creation in 2006, constituting a medium for disseminating news and political opinions on relevant topics.

Participants
We randomly selected three subjects, users of the social network Twitter, who explicitly expressed a preference for using the neuter pronouns "Elu," "El," "Ilu" or "Ile" for self-reference through the platform's tools (such as the biography, the location, and the xed tweet).To locate these subjects, an advanced search containing the referred personal pronouns was conducted, through which we intended to identify user pro les that featured one of these pronouns in the name or biography 12 .
11 "if you consider me your friend [...]," "I am terrible in the kitchen, and "how beautiful! 12It is common practice among Twitter users of the LGBTQIA+ community to add in the bio -a 160-character space for a personal introduction -the pronouns with which they prefer to be referred to, being the colloquial form of the third-person personal pronouns and possessives such as "he/she, elu/delu" are commonly used.
As criteria for subject selection, we considered (1) the participant's age, which should be over eighteen years old; (2) the account creation date should be before 2020; (3) the number of tweets made on the account, which should be signi cant: more than five thousand; and (4) the participant's posting activity should be active and frequent until the collection deadline.
The selected subjects were asked to ll in a social form whose information given follows below: Participant A: born in Varginha, MG; born in 1995 (27 years old at the time of collection); college degree; unemployed.Interested in social agendas focused on the LGBTQIA+ community and neurodiversity, actively participates in online communities focused on themes of gender identity and sexuality, as well as applies the neuter gender marking in his linguistic practices on social networks.
Participant B: born in São Paulo, capital; born in 2002 (19 years old at the time of collection); university student; unemployed and self-identi es as an agender.Interested in social agendas focused on the LGBTQIA+ community and neurodiversity.
Participant C: born in Rio de Janeiro -RJ; born in 2001 (20 years old at the time of collection); university student; unemployed and self-identi es as a cis-male.Although the participant is sympathetic to the LGBTQIA+ movement, of which he makes part, never participated in online communities focused on the debate about gender identity and sexuality; posts tweets intending to explain how gender works in the format of threads that seek to address questions about neuter language.
As gender-neuter, we considered words marked by the in ection "e" and the pronoun systems shown in Table 1 and possible variations when applying the neuter form by in ection, in which other vowels could be observed as an alternative to neutralization, such as "u"15 .Concerning the variant of other genders, only the grammatical feminine and masculine genders were considered so that other possibilities were not contemplated.
The controlled linguistic variables were Participant (A, B, and C); Lexical Item Type (random variable); Morphosyntactic Class (noun, adjective, pronoun); Topic (politics, identity, daily life, social media, and culture, a ective relationship, relationships in general, beliefs and opinions, and con icts) and Gender Marking in Generic or Speci c Reference.
The collected data were submitted to the most recent version of R software (version 4.1.2)through the RStudio interface.For statistical analysis, the data were organized in an Excel spreadsheet, converted to ".csv" format, and submitted to the functions table (), prop.table() and ggplot() for table organization, the crossing of variables, and elaboration of graphs, being the logistic regression realized through the glmer() function.

Variable Participants
We observed 3,446 occurrences of nouns, adjectives, and pronouns with masculine, feminine, or neuter gender marking -with human beings as referents -in the tweets produced by the three participants during the second half of 2020.The distribution of masculine, feminine, and neuter grammatical gender usage by participants is presented in Graph 1.
CHART 1 -Proportion of Grammatical Gender by Participant (A, B and C).
With the exception of Participant B, the distribution per participant illustrated above is characterized by a higher number of male gender occurrences, followed by the female and neuter genders.Even so, it is possible to observe that Participant A makes more expressive use of the masculine marking (74.8%), while the feminine and neuter markings present proportions closer to each other, 15.3% and 9.8%, respectively.It is possible to associate the high percentage of male labeling use for Participant A to his gender identity, which, despite falling on the non-binary spectrum and accepting the use of the neuter gender for his own reference in the social network, recognizes himself as more aligned to the male gender.In his linguistic practice on Twitter, this implies a high rate of use of the masculine grammatical gender, especially in texts whose referent is the speaker himself, as exemplified by the occurrence in (1) below.
The highlighted words are marked in the masculine form by Participant A to refer to himself, which seems to indicate -also considering the total data set -that the participant himself prefers masculine marking and uses it frequently when referring to himself, a fact that justifies the high number of masculine occurrences observed in Graph 1.
Participant B, on the other hand, presents a distribution of data by gender, different from the distributions observed in participants A and C, according to which 43% of the data consist of gender-neuter marking, while the masculine and feminine forms correspond to similar proportions, 29.3%, and 27.2%, respectively.We can notice that most occurrences of the neuter variant are observed in Participant B's data, responsible for 356 occurrences of neuter gender of the total 538 that make up the sample.
The high rate of use of gender-neuter marking for Participant B seems to be associated with their gender identities, who self-identifies as agender, and with his partner, who is frequently mentioned in the tweets that make up the sample, as seen in the example below: (2) Participant B: "mi crush disse que sou fofe." 17   The highlighted words are marked in the neuter form by Participant B when they refer to himself and their partner, which indicates, according to the data set, that Participant B prefers gender-neuter marking and uses it frequently to refer to themselves and their partner.
About Participant C dataset, we observe a distribution of data by gender similar to that of Participant A, with the prevalence of the masculine grammatical gender use followed by the feminine and neuter.The occurrences of the masculine and feminine genders, on the other hand, are closer to one another if compared to the data for Participant A, representing, respectively, 58.2% and 38.3% of the total data, with the neuter gender marking being the least used among the three participants, featuring only 3.3% for Participant C.
Similar to Participant A, Participant C uses the masculine gender to refer to themselves, which contributes to the higher number of occurrences for this gender, as depicted in Graph 1.Although this participant signals on his Twitter profile that he accepts being referred to by using neuter pronouns, they use neuter gender to mention themselves in only one of the 39 occurrences of total neuter gender uses, according to (3) below: (3) Participant C: "Nos mantem atualizades por favor." 18  As seen in ( 3), the adjective atualizado was marked in the neuter form in agreement with the first-person plural for the reference of a group of people whose gender is varied, in a generalizing reference.In specific reference, Participant C does not use neuter gender marking for their reference at any time, but only for the reference of individuals of non-binary gender identity in the second or third person.Such a communicational context (tweets in which reference is made to subjects of non-binary gender identity in the second or third person) did not occur many times in the data sample concerning Participant C, which may explain the low number of occurrences of gender-neuter observed.
Considering the three participants' data, the neuter gender marking is still less used than the other genders, even for Participant B, which presents many occurrences in the neuter form.This result shows that, although participants adopt gender-neuter markers in their tweets, the grammatical masculine and feminine genders still present a significant number of occurrences, characterizing most of the data observed in this sample.

Variable Lexical Item Type
Regarding the Lexical Item variable, 46 types marked with neuter gender were observed in the sample referring to Participant A, 108 types to Participant B, and 20 types to Participant C. Figure 1 presents the word cloud for the occurrences of Participant A. Source: Baldez (2022, p. 121) As shown in Figure 1, the words that are most often marked in the neuter form for Participant A are "Lindo" (gorgeous), occurring 23 times for the neuter variant out of a total of 96 occurrences (as in "é ume linde" 19 ), "Amigo" (friend), occurring 20 times out of 80 occurrences (as in "o que eu não faço pelus amigues20 "), and "Meu" (my), occurring 13 times out of 43 occurrences (as in "teria orgulho de ter elu como mi namorade"21 ).In addition to these, the words "Querido" (like "you are very dear"), with 7/13 occurrences, and "Fofo"22 (like "que fofinhe amg!"), with 9/48 occurrences, also stand out.Regarding the word "Querido (dear)", out of the seven data observed in the neuter form, five function as the vocative (like "sorry to worry you, queride [...]").This type is frequently used in the neuter form to refer to an interlocutor who prefers the neuter gender for self-reference (in the vocative form).
Considering the total of lexical items marked in the neuter form by Participant A, variation in the form of realization of the neuter gender was observed in some words, listed below in Table 2.As shown in Table 2, the words "Ele," "Menino," "Meu," and "Sensato" were realized in different ways in the neuter form, presenting themselves respectively in the forms "Elu/El," "Menine/Meninu," "Mi/Minhe" and "Sensate/Sensatu."Regarding realizations II of "Ele," "Menino," and "Sensato," we found that they all occurred in the same tweet, transcribed below: (4) Participant A: meninu como este Lucas é sensatu E el escreve umas coisas legais também! 24.
The above tweet is written by Participant A to a follower, who wishes to be referred to by the neuter form marked with "u" in his biography.We can observe it in the first two highlighted words in (4).This variation in the realizations of Participant A is therefore justified due to his interaction with another individual who asks for the adoption of a different neuter gender marking system for self-reference, according to which the personal pronoun in the rst-person singular is not marked by the vowel "u," being characterized only by "El", and adjectives and nouns are marked by "u" instead of "e." As for the variation of the word "Meu" between "Mi" and "Minhe," on the other hand, no evidence was observed to justify why one or the other realization was adopted.As an example, two tweets in which "Mi" and "Minhe" occur have been transcribed below: (5) Participant A: Olhem isso, mis mutuals 25 (6) Participant A: [...] Minhe lhe, eu levo TUDO para o lado pessoal 26 .
Both realizations in ( 5) and ( 6) of the neuter variant referring to the possessive pronoun are followed by a vocative in generalized reference since they address a group of people whose gender is varied.The signi cant di erence between the two examples lies in the noun following the pronoun, which in ( 5) is an English word and in ( 6) is not; however, in principle, there would be no impeding factor to the use of "Mi" together with the word "Filhe" or the use of "Minhe" together with the word "Mutual."What can be considered about this case is that it does not seem possible to deduce a rule about marking the word "My" in the neuter form from the data sample collected.
Regarding Participant B, whose data are presented in Figure 2 below, we observe the predominance of the words "Namorado (partner)" (as in "mi namorade vai ver isso comigo [...]") 27 and "Meu" (as in "pessoal eu amo mi namorade [...]") 28 , occurring in the neuter form respectively 65 times out of 71 and 61 times out of 106.Also noteworthy are the words "Ele (He)," with 37 occurrences of neuter gender from a total of 148 (as in "gosto de fazer tudo com elu") 29 , "Obrigado (Thank you)", with 20 occurrences in the neuter form out of 20 total observed occurrences (as in "Eu vou olhar!Obrigade!") 30 , 25 "Look at this, mis mutuals.""Mutual" is a Twitter slang originating from the English language.It refers to individuals who follow a social network user and are followed back.
27 "My partner will help me out". 28"Guys, I love my partner". 29"I like to do everything with elu". 30"I will look!Obrigade (Thank you)".
Source: Baldez (2022, p. 112) The high frequency of words mentioned in the neuter form seems to be associated with the topics addressed by Participant B on the social network, which mostly make comments about affective relationships (which explains the recurrence of the words "Meu (My)" and "Namorado (Partner)" in the neuter form, as well as mysticism, as can be seen in the realizations of the word "Bruxo (Witch)," frequently used by Participant B when talking about himself.About the variation of the pronoun "Meu (My)" by Participant A in the forms "Mi" and "Minhe," we observe that Participant B uses only the form "Mi," which occurred 60 times, and the plural form "Mis," only once.
An interesting fact observed about the lexical items marked in the neuter form by this participant was the use of neuter gender in the word "artista (artist)," used twice as "artiste".Despite being a word that does not present gender inflection, and thus not opposing between masculine and feminine gender, Participant B adds the "-e" mark to the word instead of the theme vowel "a" more than once, emphasizing the gender of the referred artist as observed in the occurrences "[...] façam ume artiste feliz com seus likes" 32 and "precisa de ume artiste muito talentose [...]"33 .These occurrences can be seen as an example of the use of the neuter gender to make visible the non-binary identity of the person referred to in the tweets, being a linguistic use that, even in a word in which there is no gender inflection, seeks to make evident the non-binary gender.
Figure 3 below presents the cloud of words marked with neuter gender by Participant C. According to Figure 3, the pronoun "Ele (He)" is predominant (e.g., "falaria todes elus lá pq não gosto de generalizar pelo masculino")34 marked in the neuter form, observed in 16 out of 361 occurrences.Except for "Dele" (e.g., "o que você acha delu?") 35 , which occurs twice in the gender-neuter form.All other words in Figure 3 are observed only once each in the neuter form; there is, in this case, a concentration of gender-neuter occurrences of the lexical item "Ele (He)." Additionally, no variations in gender-neuter realizations were noted for Participant C. The pronoun "Meu (My)," which showed variation for Participant A, is employed in the neuter form only once in the sample concerning Participant C, being realized as "Minhe."About the occurrences observed for the three participants, the absence of the neuter form in some demonstrative and clitic pronouns is noteworthy.More specifically, the pronouns "Esse (this)," "Lo/o (him/it)," "Desse (this)," "Nesse (this)" and "Deste (this)" did not occur in the neuter form, so it was not possible to deduce the pattern adopted for the application of the neuter marking in these words.As for the uses of neuter pronouns characterized in Table 1 (section 2), no guidelines contemplate the neuter form of the pronouns mentioned, which characterizes a limitation of the proposals of pronoun systems cited.In this sense, in pronouns where gender marking is not so intuitive, the neuter gender may not occur (as observed with the pronouns "Esse," "Lo/o," "Desse," "Nesse" and "Deste") or vary (as observed with the possessive pronoun "mi/minhe").
As for the morphosyntactic class of gender-neuter articles, there were several occurrences in which both definite and indefinite articles were gender-neuter in agreement with the noun phrase, as exemplified in "a pessoa está destinade a se apaixonar por ume assexuade"36 and "tô aliviade por não ser únique que acha esquisito," 37 for example.In the sampled occurrences, the gender-neuter seems to be identified according to the examples cited: the addition of "-e" to the root "um" is done in the case of indefinite articles, and "ê," with a circumflex accent, is used in the case of definite articles.The use of the form "Le" as a definite article, as provided in Table 1, was not observed in the data sample.

Multivariate Analysis
In order to verify how the Lexical Item and Participant variables could influence the employment of gender neutrality, a logistic regression model was developed containing the variables cited as random effects and the variables Morphosyntactic Class, Topic, and Gender Marking in Generic or Specific Reference.The results are presented in Table 2    Regarding the results obtained for the variable Topic, it can be seen that the use of the neuter gender occurs mainly in discussions involving gender identity and identity aspects, practices associated with the community of individuals of non-binary gender identity, and their supporters.In this sense, this result indicates that the marking of gender neutrality is associated with the Topic of Identity and the occurrences related to this factor are considered examples of practices in favor of discussions focused on community issues, as observed in: "Se você não é contra aros, aces, intersexos, não-bináries e xenogêneros então pq, minhe filhe, vc se diz antimogai??;" "amiguinhes bigêneros, me tirem uma dúvida?É possível ser demiboy + alguma coisa?[...]" e "[...] hétero na hora de corrigir o amigo homofóbico, de defender o amigo LGBT+, fica como?Caladinhe." 38.
As for the adjective, this factor was not observed at any time with statistical significance, favoring neuter marking only in Participant C's sample.Nevertheless, when considering the occurrences of all participants and the random variables, the adjective and the morphosyntactic classes with the highest number of occurrences in the neuter form were statistically significant.As for the result of the Generic Reference factor, it is essential to consider that gender neutrality in Specific Reference depends on a specific communicational context, which contemplates the Reference to some person of non-binary gender identity.If the participant is not interacting with or referring to a specific individual who prefers the neuter marking for selfreference, using the neuter gender in Specific Reference will not be observed.On the other hand, Generic Reference does not have such limitations of use, which means that its use can be observed in an environment in which the participant refers to people of varying or unknown gender.
38 "If you are not against aros, aces, intersexuals, non-binary and xenogenders then why, minhe filhe, you call yourself anti-mogai??"; "bigender friends, ask me a question?Is it possible to be demiboy + something?[...]" and "[...] straight when it is time to correct the homophobic friend, to defend the LGBT+ friend, what do you do?Shut up".
As for the hypothesis established for the variable Lexical Item, seen as a random effect by the statistical test, we tried to determine whether any of the 480 types that make up the variable Lexical Item favor the use of neuter gender, not disregarding the chance to occur in specific words.For this reason, the coef() function was applied to the regression model presented in Table 2, in which fitted models can be observed.According to the angular coefficients observed for the lexical items referring to the variable Lexical Item, we can verify that no word acts in favor of neuter marking, as can be seen below (Table 3), which displays the five words that were closest to a favoring context.The lexical items with the highest intercept values were, in descending order, "Namorado" (-1.43 logodds), "Querido" (-1.70 logodds), "Junto" (-1.90 logodds), "Bruxo" (-2.14 logodds), and "Lindo" (-2.24 logodds), as shown in Table 3. Considering all the types in this statistical model, the lexical item with the smallest angular coefficient, indicating the greatest disfavor upon the neuter gender, was "esse" (-4.53 logodds), a word that was not observed in the neuter form.From the reported result, we can infer that gender-neuter employment is generally disfavored, not showing optimistic estimates of any specific word.

Conclusion
Gender-neuter marking can be considered a linguistic practice associated with people of non-binary gender identity and members of the LGBTQIA+ community, as well as most of their supporters, given that its use 39 Table subtitles: Namorado -partner; querido -sweetheart; junto -together; bruxowitch; lindo -gorgeous.
represents a position of resistance to the gender binary system, expressing meanings and values representative of their users.This group can be identified in virtual environments, such as Twitter, through the practices undertaken by its members, associated with movements that promote visibility, respect, and gender equity to people of non-binary gender identity.
Through the Lexical Item variable, it was possible to observe that the variable in question did not present, as a result, positive angular coefficient values, which could indicate that a particular lexical item acts in favor of gender-neuter marking.We could also verify which words were more often marked with the neuter gender, and it was found that they vary according to the participant.The most frequently used gender-neuter words were "linde" and "amigue" for Participant A, "mi" and "namorade" for Participant B, and "elu" for Participant C.
There were also cases of variation in the realization of the neuter form in the Participant A sample, especially between the possessive pronoun "meu," realized as "mi" and "minhe," and cases of neuter gender marking in words that do not present a gender suffix in the Participant B sample observed in two occurrences containing the word "artiste".The analysis also allowed us to verify that the pronouns "Esse," "Lo/o," "Desse," "Nesse" and "Deste" did not occur in the neuter form at any time, not being possible to describe how the neuter gender is employed in these cases.
In compliance with the tenets of Variation Theory, the results show that the neuter-gender marking is a linguistic variable conditioned by the morphosyntactic class Adjective, the topics Affective Relationship and Identity, and Generic Reference Marking in words characterized by the trait [+sexed] in which the gender determination embodies the biological sex distinction through the opposition of "-a," "-o" or "-e" morphemes.The fact that the neuter gender does not occur in words whose gender marking is not attributed by "-a/-o" inflection, such as the pronouns "Esse," "Lo/o," "Desse," "Nesse" and "Deste," reinforces the conclusion.
As indicated by the result observed for the topic Identity, the three participants and their mutuals are more likely to employ the gender-neuter marking when they are engaged in a discussion that encompasses gender identity and sexuality -a practice that is associated with the community of individuals of non-binary gender identity and their supporters, according to the theoretical assumptions adopted here.The gender neuter marking is therefore understood as a tool to express individual identity and filiation to identity agendas.As a (socio)linguistic variable in Brazilian Portuguese, it highlights the protagonism of a group of people who are acting on language to expand the possibilities of the language system to express new values.The results obtained in this study, although based on the written production in a virtual environment of three participants, meet this perspective.

TABLE 3 -
Mixed effects logistic regression for the use of the neuter variant considering Morphosyntactic Class, Topic, and Generic/Specific Reference (Lexical Item and Participant as random variables) Table2shows that the intercept favors the employment of the neuter