Assessing the Mixed or Generic Feminine as an Inclusive Language Strategy

Proposed grammatical gender-neutral language practices employing rephrasing, binomials, or abbreviated double forms are unlikely to achieve general acceptance or durably modify the linguistic system given their unusual graphical features, variable treatment of speech and writing, heavy processing requirements, increased volume, and overall complexity. In contrast, use of the feminine for mixed reference is well established for female-dominated professions such as nurse, draws on established linguistic resources, and preserves correspondance between written and spoken language. We provide examples of this strategy in several languages and discuss its advantages and shortcomings.


Terminology and Issue
In numerous languages nouns are members of noun classes, which 'control' (i.e. determine) the forms of "agreement targets" (Corbett 1991, 189) such as pronouns, articles, or adjectives. Some languages (such as French, Italian, or Welsh) feature classes which are labelled 'feminine' and 'masculine', reflecting the fact that the vast majority of nouns with female and male referents are members of the corresponding noun class. Some (such as German, Lat-in, or Konkani) have a further class, typically labelled 'neuter', which contains few animate nouns. However, in general in such tripartite noun-class systems, non-animate nouns may occur not only in the neuter class, but also in the masculine and feminine class. There are other combinations and specificities, which are outside the scope of this article. Here we will concern ourselves with 'gendered' languages of this kind, which feature masculine and feminine classes only, or have a further neuter class to which non-animate nouns are not exclusively assigned. Such noun-class systems run into difficulties referencing animates without specifying their sex. This may be necessary in the case of mixed plurals, and when nouns are employed with generic value. A plurality of animates may be made up of males alone, females alone, or a mixture of the two. In many gendered languages this last case corresponds to a lexical and grammatical gap.
The binary nature of the class system imposes a choice: masculine or feminine. This choice is known as a 'resolution rule' (Corbett 1991, 261 ff). If, say, the resolution rule specifies the masculine, then the masculine becomes ambiguous, since it may reference one or more males, or a mixed group. If it is feminine, the ambiguity concerns females and mixed groups.
Many writers (e.g. Elmiger 2008;Cerquiglini 2018) have presumed that this issue is one limited to genericity, the 'generic masculine', but this is incorrect. Ultimately it concerns all nouns with mixed reference, be they plural (in which case they may be generic or specific), or singular (in which case they are necessarily generic). It is thus more accurate to refer to a 'mixed masculine': if plural the referent group is literally mixed, whether the plural holds a generic value, or refers to a specific group with masculine and feminine members; if singular and generic its semantic value is potentially mixed (it may reference a member of either sex).

'Inclusiveness' and Masculine Resolution
In most gendered languages, the resolution rule specifies the masculine. This choice has been justified in various ways. Linguistically, it has been attributed to an 'unmarked' quality attributed to the masculine (in view of its numerous systemic roles 2 ), though as a justification the argument is circular (Labrosse 1996, 30). Historically a supposed superiority of the male sex has frequently been invoked (for instance Favre de Vaugelas's (1647, 83) famous assertion of the 'nobility' of the masculine). The scope of masculine resolution has sometimes been widened, for example in purist traditions excluding certain feminine agentive nouns on nonlinguistic grounds (for French see Viennot 2014;Cerquiglini 2018).
Resistance to the masculine resolution rule has taken many forms, of a more or less radical nature. These are found in the manuals for non-sexist expression produced by individuals or state-sponsored committees. They include formulations which permit avoidance of masculine terms: 3) Potenciar y promover el rol de los agentes culturales. > el rol de las personas encargadas de la gestión cultural. 4) Se promueve la divulgación de los derechos laborales de los artistas. > para profesionales de las artes. (Guía de lenguaje inclusivo de género 2016, 5) While such strategies are accessible in their use of established linguistic resources, they nonetheless have costs. The noun groups in examples (3) and (4) increase from 8 to 15 and from 5 to 11 syllables. In (3), personas is grammatically feminine and any targets will be feminine. However in (4) profesionales is a masculine generic, and even though as a noun it is epicene, any target will return a masculine form. Hence these strategies appear appropriate for proofreading stabilized texts, but less so for spontaneous speech or correspondence, where the avoidance they deploy is either effortfully long or highly context-dependent. A second form is the binomial, which places the masculine and feminine forms in succession, rhetorically or semantically individuating the two groups, males and females. Charles de Gaulle is sometimes quoted as a precursor of such binomials (Elmiger 2008, 272); indeed Viennot (2014, 110) asserts bluntly that "il faut suivre le général et son 'Françaises, Français'" (one should follow General de Gaulle's 'Françaises, Français'). De Gaulle's phrase indicates a particular appeal to his female interlocutors, certainly justified cotextually by the se-mantic value of the following verb ("Françaises, Français, aidez-moi"), but it necessarily also invalidates the resolution rule, the masculine implicitly failing to cover both groups of referents. Binomials at least double the volume of a single gendered form (the feminine is generally longer, and the group may include a conjunction), and their use comes into conflict with the Gricean "Be brief" maxim. This is seen in De Gaulle's own practice. His first TV address after coming to power concluded with the famous phrase and its binomial, and also featured two cases of "celles et ceux", but there are also 11 examples of a simple masculine generic, including Français (5), while generic Français also crops up in another famous speech similarly concluded (6 Although not apparent in these examples, any target words controlled by these binomials will also be masculine -unless a new double form is used. A third, more recent development is the 'abbreviated double form' which uses "separation signs" (Elmiger 2008, 130-1). Where the feminine is longer than the masculine we find graphical emphasis on the morphological boundary. Brackets -enseignant(e) -have been used in this way for a long time, but their marginalizing connotations have led other forms to emerge: enseignant.e, enseignant-e, enseignant·e, enseignantE, sometimes in the absence of a morphemic boundary, as in English s/he. German has internal capitals (Binnengrossschrei-3 '[S]alariés' and also 'engagés' in this speech, are orally bivalent (Labrosse 1996, 97), and hence unmarked, despite the transcription. 4 https://fresques.ina.fr/de-gaulle/fiche-media/Gaulle00014/allocution-radio-televisee-du-27-juin-1958-prononcee-a-l-hotel-matignon.html.
5 The presence of women soldiers in the French army dates from 1938 (loi Paul-Boncour).

Mat Pires
Assessing the Mixed or Generic Feminine as an Inclusive Language Strategy

Assessing the Mixed or Generic Feminine as an Inclusive Language Strategy
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 115 Language, Gender and Hate Speech, 111-124 bung) to mark off the beginning of the suffix: LehrerIn. One shortcoming is that these forms tend to suggest the primacy of the segment left of the boundary, in general the masculine, constructing the feminine as an extension of it (Labrosse 1996, 62-3;Cerquiglini 2018, 67, fn. 2).
A final procedure is the adoption or resemanticization of individual characters; it is particularly suited to situations where the feminine substitutes, rather than expands, the masculine. Thus the atsign @ may iconically embrace O/A alternation in Spanish (7), X may stand for any letter, also in Spanish (8), and A/E alternation is served by a ligature in an unusual usage in French (9) These forms depend on the particular combination of substituted letters (French la vs le, Italian gli vs le, etc.). Both these forms and the abbreviated double forms create a peculiar orthographic complexification, making them inappropriate for learners; indeed they are in practice limited to expert language users (Manesse 2019, 115). They also resist oralization and are essentially confined to written language -texts containing such elements cannot even be read aloud.
The objections that I have indicated to these various innovations lead one to wonder whether, even within the restriction to writing, such modifications can potentially cohere into a functional system at all. As we saw in Charles de Gaulle's speeches, the effort, even for an expert user, is daunting. Several writers have pointed out cases where inclusive forms are abandoned mid-text. Manesse notes that even a schoolbook intended for 8-year-olds and promoting the use of Inclusive Writing, contained examples of masculine generics (2019, 123).
To test this functionality I analysed an issue of a magazine published by the French CGT union for its university employee members (Le lien, 196, Sept. 2018). By its very nature this is a publication (a) read by expert language users, and (b) on the left of the political spectrum, with an egalitarian and feminist outlook.
Le lien clearly has a policy of using gender-neutral language, and takes inspiration from the system set out in Haddad (2017): use of middle point as a morphological boundary, alphabetical ordering (la. le and not le.la, etc.). However the neutralization is far from systematic: only 66,2% of the applicable lexical items are given an inclusive form. Inclusivity is also highly skewed towards nominal forms: of forms remaining in the masculine, nouns account for only 26,4%, while a majority of non-nominals remain masculine: 51,9%.
The forms of linguistic inclusivity mentioned above are all distinct, and do not form an overall system for neutralizing gender in texts. However, such protocols do exist. In French, though the term 'écriture inclusive' is not necessarily associated with a specific set of recommendations, and may reference any written degendering strategy, its most influential manifestation is undoubtedly the Manuel d'écriture inclusive edited by Haddad (2017). The particular abbreviated double form proposed in this handbook inserts the middle point at the boundaries between the masculine form, the feminine suffix, and the plural suffix. This produces two distinct morphological types: one which is a letter series identical to the feminine (example 15), and one which superadds feminine endings to the masculine form (16). A final type juxtaposes the two gender-specific terms in alphabetical order (17): 15) ingénieur·e·s; sénior·e·s 16) acteur·rice·s; ceux·elles 17) du·de la; il·elle; la·le (Haddad 2017, 7, 10) This system does nothing to alleviate the problems of graphocentrism, inappropriateness for learners and general complexity. Indeed, the recourse to alphabetical order aggravates its complexity. Alphabetical order is specified where two gender-specific forms are simply juxtaposed, whether it be with the middle point (15), or as simple binomials, as in celles et ceux in (5) above. Correctly producing the forms thus requires the language user to analyse the spelling -a completely novel cognitive step in language production. As Manesse points out, it requires "une vigilance considérable pour contrôler, dans son discours oral ou écrit, le choix de mots selon l'ordre alphabétique de leur initiale, ou finale s'il s'agit de deux radicaux identiques !" (2019, 126). 7 Naturally, the recourse to the alphabet produces varying or-7 'Considerable vigilance is required, both in speech and writing, to determine word choice on the basis of the alphabetical order of the word's initial, or of its ending in the case of two identical stems'.

Assessing the Mixed or Generic Feminine as an Inclusive Language Strategy
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 117 Language, Gender and Hate Speech, 111-124 ders: les maçonnes et les maçons (f; m); but les décorateurs et les décoratrices (m; f). We might add that a majority of binomials end up with different orders according to number: 'le maçon ou la maçonne', but 'les maçonnes et les maçons'. 8 And finally, different inclusivity procedures may also generate different orders since recourse to the middle point systematically places the masculine in initial position: jaloux·se·s, but les jalouses et les jaloux.
This initial position is in fact only one of two ways the masculine is given precedence, since the system does not require a middle point between the stem and the masculine suffix: (acteur·ice and not ac-t·eur·ice) (Haddad 2017, 7). This ignoring of the stem-suffix boundary clearly constructs the masculine as the base form, from which the feminine departs, and as such sits uneasily with the overall project, as we have mentioned. Applying the alphabetical principle to middle-point forms is conceivable: it would produce maçon·onne or ja-louse·oux, forms no stranger than proposed neologisms such as agri-culteur·ice or hospitalier·ère·s. Excluding middle-point forms from the alphabetical principle ultimately seems to be a recognition of the limits of the Écriture inclusive project. By reproducing the initial position of the established mixed masculine form, one preserves a very limited readability, or recognizability at the least, with respect to the established orthographic system.
Overall, then, the complexity of Haddad's Écriture inclusive seems to limit it not only to writing but to a subset of types of written communication occurring between expert language users, and likely to be proofread in some way. Like the reformulation strategies it appears to be essentially a corrective phase of text preparation, rather than a comprehensive, workable system able to replace existing writing practice. Its use seems further to be limited to contexts valorizing the assertion of egalitarian representation of women and men.

Feminine Resolution
The attempts to replace the masculine resolution rule by associating two gender-specific lexical items either morphologically or syntactically is, then, beset by various shortcomings. I turn now to an option which does not rely on associating forms, but instead simply rewrites the resolution rule to specify the feminine gender. I will first look at examples of this phenomenon in English. English is often cited as a language little concerned with gender neutrality, since it has no grammatical gender, and an epicene third person plural form they. One corner of English where sexism may still lurk, however, is the singular generic, since at first sight any agreement target must be either he or she. There are however two well established alternatives: it for infants, and singular they (Bodine 1975;Haegeman 1981). This use of they has however often been the brunt of criticism from purists, leading some writers to opt for he: 18) Before the reader is shunted through the relatively uncharted, often even hypothetical territory which lies before him, it is perhaps only fair he be equipped with some general notion of the terrain. (Millett [1970]  Ultimately, given the existence of singular they, this English mixed feminine seems an unnecessary luxury. It is generally absent from grammars or style manuals, but is evoked by Chevalier, de Charnay and Gardelle: un she générique, à utiliser en alternance avec le traditionnel he (par exemple pour deux référents différents; ou pour une alternance d'un paragraphe à l'autre, ou d'une page web à l'autre d'un même site), qui vise à faire alors de l'humain prototypique une femme. Il est cependant peu utilisé, et restreint à certains écrits du monde universitaire ou sites pour jeunes parents (avec par exemple des antécédents du type your child), et il n'est le plus souvent pas recommandé par les guides d'usage aujourd'hui. (2017, 16) 9 The usage is considered both recent (Chevalier, de Charnay and Gardelle describe it as "emergent"; 2017, 18), and not particularly successful. I wish here to question that perspective, and suggest that while a consciously anti-sexist adoption of she in English is effective-9 'a generic she, used as an alternative to the traditional he (for instance for two different referents, or alternating from one paragraph to the next, or from one web page to the next in the same site) which aims to make the prototypical human a woman. It is however little used, and limited to certain academic texts, or sites for young parents (with antecedents such as your child) and is generally not recommended by contemporary usage manuals'.

Assessing the Mixed or Generic Feminine as an Inclusive Language Strategy
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 119 Language, Gender and Hate Speech, 111-124 ly modern, the feminine generic has a much longer pedigree, since the selection of gender may be determined pragmatically, rather than constituting a systematic and systemic feature. Consider this segment from Jane Austen: 21) "I shall be most happy to play [the harp] to you both," said Miss Crawford; "[…] for I dearly love music myself, and where the natural taste is equal the player must always be best off, for she is gratified in more ways than one […]". ([1814] 1998, 53) Here the feminine gender of the pronoun controlled by the generic 'player' is ostensibly premised on the fact that (a) the speaker is a member of the genericized class harpist, (b) she is female, and probably further (c) cultural factors mean any harpist is likely to be female. A contemporary example of such gender attribution is the following: This example may be seen as a case of a conscious feminization strategy, as in (19) and (20)  The mixed feminine does not generally figure in grammars, despite the frequency of such occurrences. The only mention in a French reference work that I have found is in Wilmet's Grammaire critique du français, where, after devoting several pages to the feminization of names of professions, the author remarks wryly: "Après cela, libre aux hommes de refuser l'installation de l'épicène féminin les infirmières" (2010, 72). 10 While Wilmet has the merit of mentioning this usage, it is nonetheless surprising to find a genuine and established refusal of masculine resolution evoked only as badinage, in a context where men wreak revenge for feminized job agentives by refusing to use it. Bickes and Brunner are similarly flippant, separating their numbered arguments in its favour into "serious" and "playfulironic" (1992( , 7-10, cited in Elmiger 2008.
Despite its scarcity in published reference works, the mixed-reference feminine is clearly recognized by speakers. Three of Elmiger's male German respondents evoke it spontaneously -thus diverging from the gender-neutral forms he has proposed for discussion. One explains that as he has already come across such a feminine, "I could also imagine people saying Bürgerinnen and Bürger being included in that […] I found it a bit unusual at the start until I realized yes why not you can turn it round", while another, a trainee nurse, evokes "one version […] we're always coming across at the school you only write the feminine form essentially... and it's spoken like that you 10 'Having said that, it falls to men to refuse to countenance the feminine epicene infirmières'.

Mat Pires
Assessing the Mixed or Generic Feminine as an Inclusive Language Strategy always address both sexes" (Elmiger 2008, 281). He has no quarrel with this usage, and indeed states that it "simplifies things considerably" (281). Similarly, Perry records a 45-year-old female observing that for French "Les métiers étaient [...], jusqu'à récemment, plutôt dotés du genre qui correspondait au sexe de la majorité des gens qui les exerçaient (une infirmière, un ambassadeur)" (2011, 319). 11 She correctly identifies the practice of number genericity, the selection of gender on the basis of the perceived majority of members of the class.
Elmiger devotes two pages to "Le féminin générique" (2008,. He begins by describing it as a "very radical" way of getting round masculine genericity, despite making no such comment on abbreviated double forms (2008, 130 ff.) incorporating obscure typographical marks which open up a chasm between spoken and written language. The radical aspect appears to be semantic: the feminine needs to be somehow redefined as mixed, in a metalinguistic note. Elmiger (2008, 143) gives examples of such notes in the feminine-generic regulations of an undefined teacher training college (28), and of the Swiss town of Köniz (29): 28) 1. Le féminin est utilisé par souci de simplification et d'entraînement pour les étudiants. 29) Für alle Funktionsbezeichnungen steht stellvertretend für beide Geschlechter die weibliche Schreibweise.
He mentions a similar attempt in the town of Wändeswil which was finally abandoned in the face of opposition. However, this failure, dating from the early 1990s, seems not to have prefigured a trend: in May 2018 the University of Neuchâtel rewrote its statutes using the generic feminine, including in their Article 1 the comment: 30) Les termes utilisés pour désigner des personnes sont pris au sens générique; ils ont à la fois valeur d'un féminin et d'un masculin.
In such documents there is of course a particular legal need to specify the unusual semantic scope of these terms. In less semantically polarized occurrences of generic infirmière, or less legalistic texts -such as the occurrences in Paris-Match or Journal de Montréal, or the English examples from London Review of Books -no such usage note is deemed necessary. However, there do seem to be some comprehension issues: when Elmiger submits to his respondents a text referring to "une cinquantaine d'infirmières (dont quatre infirmiers)" (2008,142), where the generic feminine contains a gender-specific masculine subset, most of them fail to understand the infirmiers as part of the infirmières (though at least one questions the usefulness of spec-ifying the sex at all, implying that he considers the subset to be part of the group). This clearly reflects the established specific value of the feminine. Interestingly, the established feminine generic infirmières involved in these exchanges is not presented as a precursor to more recent 'militant' uses of the feminine (cf. 2008, 142, 302).

4
Rationale for the Mixed Feminine I hope to have shown that the feminine generic avoids some of the shortcomings of inclusive writing. Its one complexity appears to be the occasional need for such metalinguistic notes, but as the study quoted above shows, already in 2011 certain language users were familiar with it. I turn now to the logic behind its adoption. The most straightforward justification for the feminine generic is number: the generic follows the numerically dominant group. My department faculty, and its students, are both female dominated. I may therefore justify writing or addressing either group in the feminine on these grounds -which are moreover exactly those behind the use of caissière or infirmière. I have used this argument in addressing a first-year undergraduate intake, and continued in subsequent correspondence, and have never received any objection. 12 There is a logic to number genericity -unlike the random choice of masculine, or indeed feminine gender. The system does not, however, have a defined outcome when the proportion of members is unknown. It also may reinforce sexual stereotyping.
It might be thought that the adoption of the feminine would be an initial step on the path to free gender alternation in the expression of mixed reference. However, such a system would remove the opposition between specific and generic altogether, making it probable that some replacement mark would emerge -a typographical feature in writing perhaps, something else in speech. This suggests that some form of number genericity, or a systematic feminine for mixed reference, as exists in certain languages, 13 are more workable options. The latter would certainly respond to the criticism motivating much of the effort towards inclusive expression: that of the visibility of women in language.
12 Clearly the dissymmetric status of teacher and student is a factor here, but I have never even received a request for clarification. On a single occasion I received an email from a student who considered himself not to be concerned by a reference in the generic feminine -and this was despite the inclusion of a usage note! 13 Corbett (1991, 220-1) gives examples of feminine for mixed reference from Maasai, Seneca, Goajiro and Dama.

Mat Pires
Assessing the Mixed or Generic Feminine as an Inclusive Language Strategy

Inclusive Forms as a Step towards a Mixed Feminine
My final point concerns ways in which the abbreviated double forms of inclusive writing may constitute a step towards feminine or number genericity, rather than being an end in themselves. Inclusive writing has certainly been more successful form of gender neutralization than the feminine generic in recent years, but in one way at least -attempts at their oralization -inclusive forms seem almost coterminous with the generic feminine. Elmiger (2008, 139) indicates three oralization strategies for German Binnengrossschreibung forms (e.g. Mu-sikerIn). The first two emphasize the suffix boundary with a slight pause or by glottalizing its initial vowel. However, a third requires no phonological innovation, since, on the observation that these forms are alphabetically feminine, it simply proposes a pronunciation identical to that of the feminine. There are limits to this strategy: the feminine of Angestellter is Angestellte, and the proposed neutralized form AngestellteR (136) is thus alphabetically masculine.
In French a great majority of inclusive agentives are alphabetically feminine (e.g. enseignant.e) and could thus be orally feminine. The main exceptional suffixes are -eur (-euse or -rice) and -aux (-ales), along with a few rare pairs like -if (-ive) or -oux (-ouse), and a handful of closed-group items (celui/celle, il/elle, le/la...). It is 'unpronounceable' forms like these -travailleur·euse·s, actif·ive·s -which stand in the way of a feminine pronunciation. One solution would be to generalize the use of alphabetical feminines and to mark their mixed status with a single (semantic) mark rather than the multiple morphological marks recommended. The suffix boundary is the obvious spot -tra-vaill·euse, act-ives -and such a system would not grant the masculine precedence as the present Écriture inclusive does. Rather than tra-vailleur·euse·s social·e·s we would have the far more approachable tra-vaill·euses soci·ales. In fact, nothing prevents the target words from being simply feminine -travaill·euses sociales -since gender is a property of the noun, and the noun is here marked (graphically) as generic; the target adjective is at no risk of being interpreted as specific.

Conclusion
I have argued that proposed gender-neutral language practices such as binomials, rephrasing, and abbreviated double forms are too complex and inherently limited to achieve general acceptance and durably modify the linguistic system. The feminine generic, on the other hand, has several advantages: it is a well established usage, it draws on existing linguistic resources, and it preserves correspondence between written and spoken language. Number genericity further constitutes a possible procedure for selecting the generic according to the situation. Finally, it is suggested that since many abbreviated double forms are alphabetically feminine, the feminine might be generalized as the mixed form, with supplementary signs reduced to a single typographical mark to indicate mixed reference. This would remove 'unpronounceable' inclusive forms, and open the way to a systematic mixed and generic feminine, a feature which would undoubtedly increase the linguistic visibility of women.